From dd3109cd89ebc350236c54db92c2c1ef7d86a843 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=9D=8E=E9=80=9A=E6=B4=B2?= Date: Fri, 27 Oct 2023 21:05:03 +0800 Subject: [PATCH] GPU: better nvidia GPU temp detection --- src/3rdparty/nvml/nvml.h | 9023 +++++++++++++++++++++------- src/3rdparty/nvml/repo.json | 4 +- src/detection/gpu/gpu_linux.c | 6 +- src/detection/gpu/gpu_windows.c | 16 +- src/detection/temps/temps_nvidia.c | 55 +- src/detection/temps/temps_nvidia.h | 2 +- 6 files changed, 6901 insertions(+), 2205 deletions(-) diff --git a/src/3rdparty/nvml/nvml.h b/src/3rdparty/nvml/nvml.h index 53bced8df..191d41da5 100644 --- a/src/3rdparty/nvml/nvml.h +++ b/src/3rdparty/nvml/nvml.h @@ -1,42 +1,42 @@ /* - * Copyright 1993-2017 NVIDIA Corporation. All rights reserved. + * Copyright 1993-2023 NVIDIA Corporation. All rights reserved. * - * NOTICE TO USER: + * NOTICE TO USER: * - * This source code is subject to NVIDIA ownership rights under U.S. and - * international Copyright laws. Users and possessors of this source code - * are hereby granted a nonexclusive, royalty-free license to use this code + * This source code is subject to NVIDIA ownership rights under U.S. and + * international Copyright laws. Users and possessors of this source code + * are hereby granted a nonexclusive, royalty-free license to use this code * in individual and commercial software. * - * NVIDIA MAKES NO REPRESENTATION ABOUT THE SUITABILITY OF THIS SOURCE - * CODE FOR ANY PURPOSE. IT IS PROVIDED "AS IS" WITHOUT EXPRESS OR - * IMPLIED WARRANTY OF ANY KIND. NVIDIA DISCLAIMS ALL WARRANTIES WITH - * REGARD TO THIS SOURCE CODE, INCLUDING ALL IMPLIED WARRANTIES OF + * NVIDIA MAKES NO REPRESENTATION ABOUT THE SUITABILITY OF THIS SOURCE + * CODE FOR ANY PURPOSE. IT IS PROVIDED "AS IS" WITHOUT EXPRESS OR + * IMPLIED WARRANTY OF ANY KIND. NVIDIA DISCLAIMS ALL WARRANTIES WITH + * REGARD TO THIS SOURCE CODE, INCLUDING ALL IMPLIED WARRANTIES OF * MERCHANTABILITY, NONINFRINGEMENT, AND FITNESS FOR A PARTICULAR PURPOSE. - * IN NO EVENT SHALL NVIDIA BE LIABLE FOR ANY SPECIAL, INDIRECT, INCIDENTAL, - * OR CONSEQUENTIAL DAMAGES, OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS - * OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE - * OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE - * OR PERFORMANCE OF THIS SOURCE CODE. + * IN NO EVENT SHALL NVIDIA BE LIABLE FOR ANY SPECIAL, INDIRECT, INCIDENTAL, + * OR CONSEQUENTIAL DAMAGES, OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS + * OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE + * OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE + * OR PERFORMANCE OF THIS SOURCE CODE. * - * U.S. Government End Users. This source code is a "commercial item" as - * that term is defined at 48 C.F.R. 2.101 (OCT 1995), consisting of - * "commercial computer software" and "commercial computer software - * documentation" as such terms are used in 48 C.F.R. 12.212 (SEPT 1995) - * and is provided to the U.S. Government only as a commercial end item. - * Consistent with 48 C.F.R.12.212 and 48 C.F.R. 227.7202-1 through - * 227.7202-4 (JUNE 1995), all U.S. Government End Users acquire the - * source code with only those rights set forth herein. + * U.S. Government End Users. This source code is a "commercial item" as + * that term is defined at 48 C.F.R. 2.101 (OCT 1995), consisting of + * "commercial computer software" and "commercial computer software + * documentation" as such terms are used in 48 C.F.R. 12.212 (SEPT 1995) + * and is provided to the U.S. Government only as a commercial end item. + * Consistent with 48 C.F.R.12.212 and 48 C.F.R. 227.7202-1 through + * 227.7202-4 (JUNE 1995), all U.S. Government End Users acquire the + * source code with only those rights set forth herein. * - * Any use of this source code in individual and commercial software must + * Any use of this source code in individual and commercial software must * include, in the user documentation and internal comments to the code, * the above Disclaimer and U.S. Government End Users Notice. */ -/* +/* NVML API Reference -The NVIDIA Management Library (NVML) is a C-based programmatic interface for monitoring and +The NVIDIA Management Library (NVML) is a C-based programmatic interface for monitoring and managing various states within NVIDIA Tesla &tm; GPUs. It is intended to be a platform for building 3rd party applications, and is also the underlying library for the NVIDIA-supported nvidia-smi tool. NVML is thread-safe so it is safe to make simultaneous NVML calls from multiple threads. @@ -52,13 +52,13 @@ Supported products: - Full Support - All Tesla products, starting with the Fermi architecture - All Quadro products, starting with the Fermi architecture - - All GRID products, starting with the Kepler architecture + - All vGPU Software products, starting with the Kepler architecture - Selected GeForce Titan products - Limited Support - All Geforce products, starting with the Fermi architecture The NVML library can be found at \%ProgramW6432\%\\"NVIDIA Corporation"\\NVSMI\\ on Windows. It is -not be added to the system path by default. To dynamically link to NVML, add this path to the PATH +not be added to the system path by default. To dynamically link to NVML, add this path to the PATH environmental variable. To dynamically load NVML, call LoadLibrary with this path. On Linux the NVML library will be found on the standard library path. For 64 bit Linux, both the 32 bit @@ -95,14 +95,37 @@ extern "C" { /** * NVML API versioning support */ -#define NVML_API_VERSION 9 -#define NVML_API_VERSION_STR "9" -#define nvmlInit nvmlInit_v2 -#define nvmlDeviceGetPciInfo nvmlDeviceGetPciInfo_v3 -#define nvmlDeviceGetCount nvmlDeviceGetCount_v2 -#define nvmlDeviceGetHandleByIndex nvmlDeviceGetHandleByIndex_v2 -#define nvmlDeviceGetHandleByPciBusId nvmlDeviceGetHandleByPciBusId_v2 -#define nvmlDeviceGetNvLinkRemotePciInfo nvmlDeviceGetNvLinkRemotePciInfo_v2 +#define NVML_API_VERSION 12 +#define NVML_API_VERSION_STR "12" +/** + * Defining NVML_NO_UNVERSIONED_FUNC_DEFS will disable "auto upgrading" of APIs. + * e.g. the user will have to call nvmlInit_v2 instead of nvmlInit. Enable this + * guard if you need to support older versions of the API + */ +#ifndef NVML_NO_UNVERSIONED_FUNC_DEFS + #define nvmlInit nvmlInit_v2 + #define nvmlDeviceGetPciInfo nvmlDeviceGetPciInfo_v3 + #define nvmlDeviceGetCount nvmlDeviceGetCount_v2 + #define nvmlDeviceGetHandleByIndex nvmlDeviceGetHandleByIndex_v2 + #define nvmlDeviceGetHandleByPciBusId nvmlDeviceGetHandleByPciBusId_v2 + #define nvmlDeviceGetNvLinkRemotePciInfo nvmlDeviceGetNvLinkRemotePciInfo_v2 + #define nvmlDeviceRemoveGpu nvmlDeviceRemoveGpu_v2 + #define nvmlDeviceGetGridLicensableFeatures nvmlDeviceGetGridLicensableFeatures_v4 + #define nvmlEventSetWait nvmlEventSetWait_v2 + #define nvmlDeviceGetAttributes nvmlDeviceGetAttributes_v2 + #define nvmlComputeInstanceGetInfo nvmlComputeInstanceGetInfo_v2 + #define nvmlDeviceGetComputeRunningProcesses nvmlDeviceGetComputeRunningProcesses_v3 + #define nvmlDeviceGetGraphicsRunningProcesses nvmlDeviceGetGraphicsRunningProcesses_v3 + #define nvmlDeviceGetMPSComputeRunningProcesses nvmlDeviceGetMPSComputeRunningProcesses_v3 + #define nvmlBlacklistDeviceInfo_t nvmlExcludedDeviceInfo_t + #define nvmlGetBlacklistDeviceCount nvmlGetExcludedDeviceCount + #define nvmlGetBlacklistDeviceInfoByIndex nvmlGetExcludedDeviceInfoByIndex + #define nvmlDeviceGetGpuInstancePossiblePlacements nvmlDeviceGetGpuInstancePossiblePlacements_v2 + #define nvmlVgpuInstanceGetLicenseInfo nvmlVgpuInstanceGetLicenseInfo_v2 +#endif // #ifndef NVML_NO_UNVERSIONED_FUNC_DEFS + +#define NVML_STRUCT_VERSION(data, ver) (unsigned int)(sizeof(nvml ## data ## _v ## ver ## _t) | \ + (ver << 24U)) /***************************************************************************************************/ /** @defgroup nvmlDeviceStructs Device Structs @@ -147,13 +170,30 @@ typedef struct nvmlPciInfo_st char busId[NVML_DEVICE_PCI_BUS_ID_BUFFER_SIZE]; //!< The tuple domain:bus:device.function PCI identifier (& NULL terminator) } nvmlPciInfo_t; +/** + * PCI format string for ::busIdLegacy + */ +#define NVML_DEVICE_PCI_BUS_ID_LEGACY_FMT "%04X:%02X:%02X.0" + +/** + * PCI format string for ::busId + */ +#define NVML_DEVICE_PCI_BUS_ID_FMT "%08X:%02X:%02X.0" + +/** + * Utility macro for filling the pci bus id format from a nvmlPciInfo_t + */ +#define NVML_DEVICE_PCI_BUS_ID_FMT_ARGS(pciInfo) (pciInfo)->domain, \ + (pciInfo)->bus, \ + (pciInfo)->device + /** * Detailed ECC error counts for a device. * * @deprecated Different GPU families can have different memory error counters * See \ref nvmlDeviceGetMemoryErrorCounter */ -typedef struct nvmlEccErrorCounts_st +typedef struct nvmlEccErrorCounts_st { unsigned long long l1Cache; //!< L1 cache errors unsigned long long l2Cache; //!< L2 cache errors @@ -161,26 +201,44 @@ typedef struct nvmlEccErrorCounts_st unsigned long long registerFile; //!< Register file errors } nvmlEccErrorCounts_t; -/** +/** * Utilization information for a device. * Each sample period may be between 1 second and 1/6 second, depending on the product being queried. */ -typedef struct nvmlUtilization_st +typedef struct nvmlUtilization_st { unsigned int gpu; //!< Percent of time over the past sample period during which one or more kernels was executing on the GPU unsigned int memory; //!< Percent of time over the past sample period during which global (device) memory was being read or written } nvmlUtilization_t; -/** - * Memory allocation information for a device. +/** + * Memory allocation information for a device (v1). + * The total amount is equal to the sum of the amounts of free and used memory. */ -typedef struct nvmlMemory_st +typedef struct nvmlMemory_st { - unsigned long long total; //!< Total installed FB memory (in bytes) - unsigned long long free; //!< Unallocated FB memory (in bytes) - unsigned long long used; //!< Allocated FB memory (in bytes). Note that the driver/GPU always sets aside a small amount of memory for bookkeeping + unsigned long long total; //!< Total physical device memory (in bytes) + unsigned long long free; //!< Unallocated device memory (in bytes) + unsigned long long used; //!< Sum of Reserved and Allocated device memory (in bytes). + //!< Note that the driver/GPU always sets aside a small amount of memory for bookkeeping } nvmlMemory_t; +/** + * Memory allocation information for a device (v2). + * + * Version 2 adds versioning for the struct and the amount of system-reserved memory as an output. + */ +typedef struct nvmlMemory_v2_st +{ + unsigned int version; //!< Structure format version (must be 2) + unsigned long long total; //!< Total physical device memory (in bytes) + unsigned long long reserved; //!< Device memory (in bytes) reserved for system use (driver or firmware) + unsigned long long free; //!< Unallocated device memory (in bytes) + unsigned long long used; //!< Allocated device memory (in bytes). +} nvmlMemory_v2_t; + +#define nvmlMemory_v2 NVML_STRUCT_VERSION(Memory, 2) + /** * BAR1 Memory allocation Information for a device */ @@ -191,16 +249,104 @@ typedef struct nvmlBAR1Memory_st unsigned long long bar1Used; //!< Allocated Used Memory (in bytes) }nvmlBAR1Memory_t; +/** + * Information about running compute processes on the GPU, legacy version + * for older versions of the API. + */ +typedef struct nvmlProcessInfo_v1_st +{ + unsigned int pid; //!< Process ID + unsigned long long usedGpuMemory; //!< Amount of used GPU memory in bytes. + //! Under WDDM, \ref NVML_VALUE_NOT_AVAILABLE is always reported + //! because Windows KMD manages all the memory and not the NVIDIA driver +} nvmlProcessInfo_v1_t; + /** * Information about running compute processes on the GPU */ -typedef struct nvmlProcessInfo_st +typedef struct nvmlProcessInfo_v2_st { - unsigned int pid; //!< Process ID - unsigned long long usedGpuMemory; //!< Amount of used GPU memory in bytes. - //! Under WDDM, \ref NVML_VALUE_NOT_AVAILABLE is always reported - //! because Windows KMD manages all the memory and not the NVIDIA driver -} nvmlProcessInfo_t; + unsigned int pid; //!< Process ID + unsigned long long usedGpuMemory; //!< Amount of used GPU memory in bytes. + //! Under WDDM, \ref NVML_VALUE_NOT_AVAILABLE is always reported + //! because Windows KMD manages all the memory and not the NVIDIA driver + unsigned int gpuInstanceId; //!< If MIG is enabled, stores a valid GPU instance ID. gpuInstanceId is set to + // 0xFFFFFFFF otherwise. + unsigned int computeInstanceId; //!< If MIG is enabled, stores a valid compute instance ID. computeInstanceId is set to + // 0xFFFFFFFF otherwise. +} nvmlProcessInfo_v2_t, nvmlProcessInfo_t; + +/** + * Information about running process on the GPU with protected memory + */ +typedef struct +{ + unsigned int pid; //!< Process ID + unsigned long long usedGpuMemory; //!< Amount of used GPU memory in bytes. + //! Under WDDM, \ref NVML_VALUE_NOT_AVAILABLE is always reported + //! because Windows KMD manages all the memory and not the NVIDIA driver + unsigned int gpuInstanceId; //!< If MIG is enabled, stores a valid GPU instance ID. gpuInstanceId is + // set to 0xFFFFFFFF otherwise. + unsigned int computeInstanceId; //!< If MIG is enabled, stores a valid compute instance ID. computeInstanceId + // is set to 0xFFFFFFFF otherwise. + unsigned long long usedGpuCcProtectedMemory; //!< Amount of used GPU conf compute protected memory in bytes. +} nvmlProcessDetail_v1_t; + +/** + * Information about all running processes on the GPU for the given mode + */ +typedef struct +{ + unsigned int version; //!< Struct version, MUST be nvmlProcessDetailList_v1 + unsigned int mode; //!< Process mode(Compute/Graphics/MPSCompute) + unsigned int numProcArrayEntries; //!< Number of process entries in procArray + nvmlProcessDetail_v1_t *procArray; //!< Process array +} nvmlProcessDetailList_v1_t; + +typedef nvmlProcessDetailList_v1_t nvmlProcessDetailList_t; + +/** + * nvmlProcessDetailList version + */ +#define nvmlProcessDetailList_v1 NVML_STRUCT_VERSION(ProcessDetailList, 1) + +typedef struct nvmlDeviceAttributes_st +{ + unsigned int multiprocessorCount; //!< Streaming Multiprocessor count + unsigned int sharedCopyEngineCount; //!< Shared Copy Engine count + unsigned int sharedDecoderCount; //!< Shared Decoder Engine count + unsigned int sharedEncoderCount; //!< Shared Encoder Engine count + unsigned int sharedJpegCount; //!< Shared JPEG Engine count + unsigned int sharedOfaCount; //!< Shared OFA Engine count + unsigned int gpuInstanceSliceCount; //!< GPU instance slice count + unsigned int computeInstanceSliceCount; //!< Compute instance slice count + unsigned long long memorySizeMB; //!< Device memory size (in MiB) +} nvmlDeviceAttributes_t; + +/** + * C2C Mode information for a device + */ +typedef struct +{ + unsigned int isC2cEnabled; +} nvmlC2cModeInfo_v1_t; + +#define nvmlC2cModeInfo_v1 NVML_STRUCT_VERSION(C2cModeInfo, 1) + +/** + * Possible values that classify the remap availability for each bank. The max + * field will contain the number of banks that have maximum remap availability + * (all reserved rows are available). None means that there are no reserved + * rows available. + */ +typedef struct nvmlRowRemapperHistogramValues_st +{ + unsigned int max; + unsigned int high; + unsigned int partial; + unsigned int low; + unsigned int none; +} nvmlRowRemapperHistogramValues_t; /** * Enum to represent type of bridge chip @@ -208,13 +354,13 @@ typedef struct nvmlProcessInfo_st typedef enum nvmlBridgeChipType_enum { NVML_BRIDGE_CHIP_PLX = 0, - NVML_BRIDGE_CHIP_BRO4 = 1 + NVML_BRIDGE_CHIP_BRO4 = 1 }nvmlBridgeChipType_t; /** - * Maximum number of NvLink links supported + * Maximum number of NvLink links supported */ -#define NVML_NVLINK_MAX_LINKS 6 +#define NVML_NVLINK_MAX_LINKS 18 /** * Enum to represent the NvLink utilization counter packet units @@ -224,7 +370,7 @@ typedef enum nvmlNvLinkUtilizationCountUnits_enum NVML_NVLINK_COUNTER_UNIT_CYCLES = 0, // count by cycles NVML_NVLINK_COUNTER_UNIT_PACKETS = 1, // count by packets NVML_NVLINK_COUNTER_UNIT_BYTES = 2, // count by bytes - + NVML_NVLINK_COUNTER_UNIT_RESERVED = 3, // count reserved for internal use // this must be last NVML_NVLINK_COUNTER_UNIT_COUNT } nvmlNvLinkUtilizationCountUnits_t; @@ -234,7 +380,7 @@ typedef enum nvmlNvLinkUtilizationCountUnits_enum * ** this is ONLY applicable with the units as packets or bytes * ** as specified in \a nvmlNvLinkUtilizationCountUnits_t * ** all packet filter descriptions are target GPU centric - * ** these can be "OR'd" together + * ** these can be "OR'd" together */ typedef enum nvmlNvLinkUtilizationCountPktTypes_enum { @@ -249,7 +395,7 @@ typedef enum nvmlNvLinkUtilizationCountPktTypes_enum NVML_NVLINK_COUNTER_PKTFILTER_ALL = 0xFF // all packets } nvmlNvLinkUtilizationCountPktTypes_t; -/** +/** * Struct to define the NVLINK counter controls */ typedef struct nvmlNvLinkUtilizationControl_st @@ -282,11 +428,23 @@ typedef enum nvmlNvLinkErrorCounter_enum NVML_NVLINK_ERROR_DL_RECOVERY = 1, // Data link transmit recovery error counter NVML_NVLINK_ERROR_DL_CRC_FLIT = 2, // Data link receive flow control digit CRC error counter NVML_NVLINK_ERROR_DL_CRC_DATA = 3, // Data link receive data CRC error counter + NVML_NVLINK_ERROR_DL_ECC_DATA = 4, // Data link receive data ECC error counter // this must be last NVML_NVLINK_ERROR_COUNT } nvmlNvLinkErrorCounter_t; +/** + * Enum to represent NvLink's remote device type + */ +typedef enum nvmlIntNvLinkDeviceType_enum +{ + NVML_NVLINK_DEVICE_TYPE_GPU = 0x00, + NVML_NVLINK_DEVICE_TYPE_IBMNPU = 0x01, + NVML_NVLINK_DEVICE_TYPE_SWITCH = 0x02, + NVML_NVLINK_DEVICE_TYPE_UNKNOWN = 0xFF +} nvmlIntNvLinkDeviceType_t; + /** * Represents level relationships within a system between two GPUs * The enums are spaced to allow for future relationships @@ -298,7 +456,7 @@ typedef enum nvmlGpuLevel_enum NVML_TOPOLOGY_MULTIPLE = 20, // all devices that need not traverse a host bridge NVML_TOPOLOGY_HOSTBRIDGE = 30, // all devices that are connected to the same host bridge NVML_TOPOLOGY_NODE = 40, // all devices that are connected to the same NUMA node but possibly multiple host bridges - NVML_TOPOLOGY_SYSTEM = 50, // all devices in the system + NVML_TOPOLOGY_SYSTEM = 50 // all devices in the system // there is purposefully no COUNT here because of the need for spacing above } nvmlGpuTopologyLevel_t; @@ -311,6 +469,7 @@ typedef enum nvmlGpuP2PStatus_enum { NVML_P2P_STATUS_OK = 0, NVML_P2P_STATUS_CHIPSET_NOT_SUPPORED, + NVML_P2P_STATUS_CHIPSET_NOT_SUPPORTED = NVML_P2P_STATUS_CHIPSET_NOT_SUPPORED, NVML_P2P_STATUS_GPU_NOT_SUPPORTED, NVML_P2P_STATUS_IOH_TOPOLOGY_NOT_SUPPORTED, NVML_P2P_STATUS_DISABLED_BY_REGKEY, @@ -340,12 +499,12 @@ typedef enum nvmlGpuP2PCapsIndex_enum */ typedef struct nvmlBridgeChipInfo_st { - nvmlBridgeChipType_t type; //!< Type of Bridge Chip + nvmlBridgeChipType_t type; //!< Type of Bridge Chip unsigned int fwVersion; //!< Firmware Version. 0=Version is unavailable }nvmlBridgeChipInfo_t; /** - * This structure stores the complete Hierarchy of the Bridge Chip within the board. The immediate + * This structure stores the complete Hierarchy of the Bridge Chip within the board. The immediate * bridge is stored at index 0 of bridgeInfoList, parent to immediate bridge is at index 1 and so forth. */ typedef struct nvmlBridgeChipHierarchy_st @@ -363,12 +522,13 @@ typedef enum nvmlSamplingType_enum NVML_GPU_UTILIZATION_SAMPLES = 1, //!< To represent percent of time during which one or more kernels was executing on the GPU NVML_MEMORY_UTILIZATION_SAMPLES = 2, //!< To represent percent of time during which global (device) memory was being read or written NVML_ENC_UTILIZATION_SAMPLES = 3, //!< To represent percent of time during which NVENC remains busy - NVML_DEC_UTILIZATION_SAMPLES = 4, //!< To represent percent of time during which NVDEC remains busy + NVML_DEC_UTILIZATION_SAMPLES = 4, //!< To represent percent of time during which NVDEC remains busy NVML_PROCESSOR_CLK_SAMPLES = 5, //!< To represent processor clock samples NVML_MEMORY_CLK_SAMPLES = 6, //!< To represent memory clock samples - + NVML_MODULE_POWER_SAMPLES = 7, //!< To represent module power samples for total module starting Grace Hopper + // Keep this last - NVML_SAMPLINGTYPE_COUNT + NVML_SAMPLINGTYPE_COUNT }nvmlSamplingType_t; /** @@ -378,7 +538,7 @@ typedef enum nvmlPcieUtilCounter_enum { NVML_PCIE_UTIL_TX_BYTES = 0, // 1KB granularity NVML_PCIE_UTIL_RX_BYTES = 1, // 1KB granularity - + // Keep this last NVML_PCIE_UTIL_COUNT } nvmlPcieUtilCounter_t; @@ -386,13 +546,14 @@ typedef enum nvmlPcieUtilCounter_enum /** * Represents the type for sample value returned */ -typedef enum nvmlValueType_enum +typedef enum nvmlValueType_enum { NVML_VALUE_TYPE_DOUBLE = 0, NVML_VALUE_TYPE_UNSIGNED_INT = 1, NVML_VALUE_TYPE_UNSIGNED_LONG = 2, NVML_VALUE_TYPE_UNSIGNED_LONG_LONG = 3, NVML_VALUE_TYPE_SIGNED_LONG_LONG = 4, + NVML_VALUE_TYPE_SIGNED_INT = 5, // Keep this last NVML_VALUE_TYPE_COUNT @@ -405,6 +566,7 @@ typedef enum nvmlValueType_enum typedef union nvmlValue_st { double dVal; //!< If the value is double + int siVal; //!< If the value is signed int unsigned int uiVal; //!< If the value is unsigned int unsigned long ulVal; //!< If the value is unsigned long unsigned long long ullVal; //!< If the value is unsigned long long @@ -414,14 +576,14 @@ typedef union nvmlValue_st /** * Information for Sample */ -typedef struct nvmlSample_st +typedef struct nvmlSample_st { unsigned long long timeStamp; //!< CPU Timestamp in microseconds - nvmlValue_t sampleValue; //!< Sample Value + nvmlValue_t sampleValue; //!< Sample Value }nvmlSample_t; /** - * Represents type of perf policy for which violation times can be queried + * Represents type of perf policy for which violation times can be queried */ typedef enum nvmlPerfPolicyType_enum { @@ -448,6 +610,60 @@ typedef struct nvmlViolationTime_st unsigned long long violationTime; //!< violationTime in Nanoseconds }nvmlViolationTime_t; +#define NVML_MAX_THERMAL_SENSORS_PER_GPU 3 + +typedef enum +{ + NVML_THERMAL_TARGET_NONE = 0, + NVML_THERMAL_TARGET_GPU = 1, //!< GPU core temperature requires NvPhysicalGpuHandle + NVML_THERMAL_TARGET_MEMORY = 2, //!< GPU memory temperature requires NvPhysicalGpuHandle + NVML_THERMAL_TARGET_POWER_SUPPLY = 4, //!< GPU power supply temperature requires NvPhysicalGpuHandle + NVML_THERMAL_TARGET_BOARD = 8, //!< GPU board ambient temperature requires NvPhysicalGpuHandle + NVML_THERMAL_TARGET_VCD_BOARD = 9, //!< Visual Computing Device Board temperature requires NvVisualComputingDeviceHandle + NVML_THERMAL_TARGET_VCD_INLET = 10, //!< Visual Computing Device Inlet temperature requires NvVisualComputingDeviceHandle + NVML_THERMAL_TARGET_VCD_OUTLET = 11, //!< Visual Computing Device Outlet temperature requires NvVisualComputingDeviceHandle + + NVML_THERMAL_TARGET_ALL = 15, + NVML_THERMAL_TARGET_UNKNOWN = -1, +} nvmlThermalTarget_t; + +typedef enum +{ + NVML_THERMAL_CONTROLLER_NONE = 0, + NVML_THERMAL_CONTROLLER_GPU_INTERNAL, + NVML_THERMAL_CONTROLLER_ADM1032, + NVML_THERMAL_CONTROLLER_ADT7461, + NVML_THERMAL_CONTROLLER_MAX6649, + NVML_THERMAL_CONTROLLER_MAX1617, + NVML_THERMAL_CONTROLLER_LM99, + NVML_THERMAL_CONTROLLER_LM89, + NVML_THERMAL_CONTROLLER_LM64, + NVML_THERMAL_CONTROLLER_G781, + NVML_THERMAL_CONTROLLER_ADT7473, + NVML_THERMAL_CONTROLLER_SBMAX6649, + NVML_THERMAL_CONTROLLER_VBIOSEVT, + NVML_THERMAL_CONTROLLER_OS, + NVML_THERMAL_CONTROLLER_NVSYSCON_CANOAS, + NVML_THERMAL_CONTROLLER_NVSYSCON_E551, + NVML_THERMAL_CONTROLLER_MAX6649R, + NVML_THERMAL_CONTROLLER_ADT7473S, + NVML_THERMAL_CONTROLLER_UNKNOWN = -1, +} nvmlThermalController_t; + +typedef struct +{ + unsigned int count; + struct + { + nvmlThermalController_t controller; + int defaultMinTemp; + int defaultMaxTemp; + int currentTemp; + nvmlThermalTarget_t target; + } sensor[NVML_MAX_THERMAL_SENSORS_PER_GPU]; + +} nvmlGpuThermalSettings_t; + /** @} */ /***************************************************************************************************/ @@ -456,31 +672,43 @@ typedef struct nvmlViolationTime_st */ /***************************************************************************************************/ -/** - * Generic enable/disable enum. +/** + * Generic enable/disable enum. */ -typedef enum nvmlEnableState_enum +typedef enum nvmlEnableState_enum { - NVML_FEATURE_DISABLED = 0, //!< Feature disabled + NVML_FEATURE_DISABLED = 0, //!< Feature disabled NVML_FEATURE_ENABLED = 1 //!< Feature enabled } nvmlEnableState_t; //! Generic flag used to specify the default behavior of some functions. See description of particular functions for details. -#define nvmlFlagDefault 0x00 +#define nvmlFlagDefault 0x00 //! Generic flag used to force some behavior. See description of particular functions for details. -#define nvmlFlagForce 0x01 +#define nvmlFlagForce 0x01 /** * * The Brand of the GPU * */ typedef enum nvmlBrandType_enum { - NVML_BRAND_UNKNOWN = 0, - NVML_BRAND_QUADRO = 1, - NVML_BRAND_TESLA = 2, - NVML_BRAND_NVS = 3, - NVML_BRAND_GRID = 4, - NVML_BRAND_GEFORCE = 5, + NVML_BRAND_UNKNOWN = 0, + NVML_BRAND_QUADRO = 1, + NVML_BRAND_TESLA = 2, + NVML_BRAND_NVS = 3, + NVML_BRAND_GRID = 4, // Deprecated from API reporting. Keeping definition for backward compatibility. + NVML_BRAND_GEFORCE = 5, + NVML_BRAND_TITAN = 6, + NVML_BRAND_NVIDIA_VAPPS = 7, // NVIDIA Virtual Applications + NVML_BRAND_NVIDIA_VPC = 8, // NVIDIA Virtual PC + NVML_BRAND_NVIDIA_VCS = 9, // NVIDIA Virtual Compute Server + NVML_BRAND_NVIDIA_VWS = 10, // NVIDIA RTX Virtual Workstation + NVML_BRAND_NVIDIA_CLOUD_GAMING = 11, // NVIDIA Cloud Gaming + NVML_BRAND_NVIDIA_VGAMING = NVML_BRAND_NVIDIA_CLOUD_GAMING, // Deprecated from API reporting. Keeping definition for backward compatibility. + NVML_BRAND_QUADRO_RTX = 12, + NVML_BRAND_NVIDIA_RTX = 13, + NVML_BRAND_NVIDIA = 14, + NVML_BRAND_GEFORCE_RTX = 15, // Unused + NVML_BRAND_TITAN_RTX = 16, // Unused // Keep this last NVML_BRAND_COUNT @@ -491,45 +719,94 @@ typedef enum nvmlBrandType_enum */ typedef enum nvmlTemperatureThresholds_enum { - NVML_TEMPERATURE_THRESHOLD_SHUTDOWN = 0, // Temperature at which the GPU will shut down - // for HW protection - NVML_TEMPERATURE_THRESHOLD_SLOWDOWN = 1, // Temperature at which the GPU will begin HW slowdown - NVML_TEMPERATURE_THRESHOLD_MEM_MAX = 2, // Memory Temperature at which the GPU will begin SW slowdown - NVML_TEMPERATURE_THRESHOLD_GPU_MAX = 3, // GPU Temperature at which the GPU can be throttled below base clock + NVML_TEMPERATURE_THRESHOLD_SHUTDOWN = 0, // Temperature at which the GPU will + // shut down for HW protection + NVML_TEMPERATURE_THRESHOLD_SLOWDOWN = 1, // Temperature at which the GPU will + // begin HW slowdown + NVML_TEMPERATURE_THRESHOLD_MEM_MAX = 2, // Memory Temperature at which the GPU will + // begin SW slowdown + NVML_TEMPERATURE_THRESHOLD_GPU_MAX = 3, // GPU Temperature at which the GPU + // can be throttled below base clock + NVML_TEMPERATURE_THRESHOLD_ACOUSTIC_MIN = 4, // Minimum GPU Temperature that can be + // set as acoustic threshold + NVML_TEMPERATURE_THRESHOLD_ACOUSTIC_CURR = 5, // Current temperature that is set as + // acoustic threshold. + NVML_TEMPERATURE_THRESHOLD_ACOUSTIC_MAX = 6, // Maximum GPU temperature that can be + // set as acoustic threshold. // Keep this last NVML_TEMPERATURE_THRESHOLD_COUNT } nvmlTemperatureThresholds_t; -/** - * Temperature sensors. +/** + * Temperature sensors. */ -typedef enum nvmlTemperatureSensors_enum +typedef enum nvmlTemperatureSensors_enum { NVML_TEMPERATURE_GPU = 0, //!< Temperature sensor for the GPU die - + // Keep this last NVML_TEMPERATURE_COUNT } nvmlTemperatureSensors_t; -/** - * Compute mode. +/** + * Compute mode. * * NVML_COMPUTEMODE_EXCLUSIVE_PROCESS was added in CUDA 4.0. - * Earlier CUDA versions supported a single exclusive mode, + * Earlier CUDA versions supported a single exclusive mode, * which is equivalent to NVML_COMPUTEMODE_EXCLUSIVE_THREAD in CUDA 4.0 and beyond. */ -typedef enum nvmlComputeMode_enum +typedef enum nvmlComputeMode_enum { NVML_COMPUTEMODE_DEFAULT = 0, //!< Default compute mode -- multiple contexts per device NVML_COMPUTEMODE_EXCLUSIVE_THREAD = 1, //!< Support Removed NVML_COMPUTEMODE_PROHIBITED = 2, //!< Compute-prohibited mode -- no contexts per device NVML_COMPUTEMODE_EXCLUSIVE_PROCESS = 3, //!< Compute-exclusive-process mode -- only one context per device, usable from multiple threads at a time - + // Keep this last NVML_COMPUTEMODE_COUNT } nvmlComputeMode_t; -/** +/** + * Max Clock Monitors available + */ +#define MAX_CLK_DOMAINS 32 + +/** + * Clock Monitor error types + */ +typedef struct nvmlClkMonFaultInfo_struct { + /** + * The Domain which faulted + */ + unsigned int clkApiDomain; + + /** + * Faults Information + */ + unsigned int clkDomainFaultMask; +} nvmlClkMonFaultInfo_t; + +/** + * Clock Monitor Status + */ +typedef struct nvmlClkMonStatus_status { + /** + * Fault status Indicator + */ + unsigned int bGlobalStatus; + + /** + * Total faulted domain numbers + */ + unsigned int clkMonListSize; + + /** + * The fault Information structure + */ + nvmlClkMonFaultInfo_t clkMonList[MAX_CLK_DOMAINS]; +} nvmlClkMonStatus_t; + +/** * ECC bit types. * * @deprecated See \ref nvmlMemoryErrorType_t for a more flexible type @@ -557,56 +834,56 @@ typedef enum nvmlMemoryErrorType_enum { /** * A memory error that was corrected - * + * * For ECC errors, these are single bit errors * For Texture memory, these are errors fixed by resend */ NVML_MEMORY_ERROR_TYPE_CORRECTED = 0, /** * A memory error that was not corrected - * + * * For ECC errors, these are double bit errors * For Texture memory, these are errors where the resend fails */ NVML_MEMORY_ERROR_TYPE_UNCORRECTED = 1, - - + + // Keep this last NVML_MEMORY_ERROR_TYPE_COUNT //!< Count of memory error types } nvmlMemoryErrorType_t; -/** - * ECC counter types. +/** + * ECC counter types. * * Note: Volatile counts are reset each time the driver loads. On Windows this is once per boot. On Linux this can be more frequent. - * On Linux the driver unloads when no active clients exist. If persistence mode is enabled or there is always a driver + * On Linux the driver unloads when no active clients exist. If persistence mode is enabled or there is always a driver * client active (e.g. X11), then Linux also sees per-boot behavior. If not, volatile counts are reset each time a compute app * is run. */ -typedef enum nvmlEccCounterType_enum +typedef enum nvmlEccCounterType_enum { NVML_VOLATILE_ECC = 0, //!< Volatile counts are reset each time the driver loads. NVML_AGGREGATE_ECC = 1, //!< Aggregate counts persist across reboots (i.e. for the lifetime of the device) - + // Keep this last NVML_ECC_COUNTER_TYPE_COUNT //!< Count of memory counter types } nvmlEccCounterType_t; -/** - * Clock types. - * +/** + * Clock types. + * * All speeds are in Mhz. */ -typedef enum nvmlClockType_enum +typedef enum nvmlClockType_enum { NVML_CLOCK_GRAPHICS = 0, //!< Graphics clock domain NVML_CLOCK_SM = 1, //!< SM clock domain NVML_CLOCK_MEM = 2, //!< Memory clock domain NVML_CLOCK_VIDEO = 3, //!< Video encoder/decoder clock domain - + // Keep this last - NVML_CLOCK_COUNT //usedGpuMemory is not supported - - unsigned long long time; //!< Amount of time in ms during which the compute context was active. The time is reported as 0 if + + unsigned long long time; //!< Amount of time in ms during which the compute context was active. The time is reported as 0 if //!< the process is not terminated - + unsigned long long startTime; //!< CPU Timestamp in usec representing start time for the process - + unsigned int isRunning; //!< Flag to represent if the process is running (1 for running, 0 for terminated) unsigned int reserved[5]; //!< Reserved for future use @@ -1259,145 +2237,24 @@ typedef struct nvmlAccountingStats_st { /** @} */ -/***************************************************************************************************/ -/** @defgroup nvmlVgpuConstants Vgpu Constants - * @{ - */ -/***************************************************************************************************/ - -/** - * Buffer size guaranteed to be large enough for \ref nvmlVgpuTypeGetLicense - */ -#define NVML_GRID_LICENSE_BUFFER_SIZE 128 - -#define NVML_VGPU_NAME_BUFFER_SIZE 64 - -#define NVML_MAX_VGPU_TYPES_PER_PGPU 17 - -#define NVML_MAX_VGPU_INSTANCES_PER_PGPU 24 - -#define NVML_GRID_LICENSE_FEATURE_MAX_COUNT 3 - -#define NVML_GRID_LICENSE_INFO_MAX_LENGTH 128 - -/** @} */ - -/***************************************************************************************************/ -/** @defgroup nvmlVgpuEnum Vgpu Enum - * @{ - */ -/***************************************************************************************************/ - -/*! - * Types of VM identifiers - */ -typedef enum nvmlVgpuVmIdType { - NVML_VGPU_VM_ID_DOMAIN_ID = 0, //!< VM ID represents DOMAIN ID - NVML_VGPU_VM_ID_UUID = 1, //!< VM ID represents UUID -} nvmlVgpuVmIdType_t; - -// vGPU GUEST info state. -typedef enum nvmlVgpuGuestInfoState_enum -{ - NVML_VGPU_INSTANCE_GUEST_INFO_STATE_UNINITIALIZED = 0, //= 0 and < \a unitCount * @param unit Reference in which to return the unit handle - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a unit has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a index is invalid or \a unit is NULL @@ -1699,8 +2841,8 @@ nvmlReturn_t DECLDIR nvmlUnitGetHandleByIndex(unsigned int index, nvmlUnit_t *un * * @param unit The identifier of the target unit * @param info Reference in which to return the unit information - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a info has been populated * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a unit is invalid or \a info is NULL @@ -1716,14 +2858,14 @@ nvmlReturn_t DECLDIR nvmlUnitGetUnitInfo(nvmlUnit_t unit, nvmlUnitInfo_t *info); * * @param unit The identifier of the target unit * @param state Reference in which to return the current LED state - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a state has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a unit is invalid or \a state is NULL * - \ref NVML_ERROR_NOT_SUPPORTED if this is not an S-class product * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * + * * @see nvmlUnitSetLedState() */ nvmlReturn_t DECLDIR nvmlUnitGetLedState(nvmlUnit_t unit, nvmlLedState_t *state); @@ -1737,8 +2879,8 @@ nvmlReturn_t DECLDIR nvmlUnitGetLedState(nvmlUnit_t unit, nvmlLedState_t *state) * * @param unit The identifier of the target unit * @param psu Reference in which to return the PSU information - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a psu has been populated * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a unit is invalid or \a psu is NULL @@ -1752,14 +2894,14 @@ nvmlReturn_t DECLDIR nvmlUnitGetPsuInfo(nvmlUnit_t unit, nvmlPSUInfo_t *psu); * * For S-class products. * - * Depending on the product, readings may be available for intake (type=0), + * Depending on the product, readings may be available for intake (type=0), * exhaust (type=1) and board (type=2). * * @param unit The identifier of the target unit * @param type The type of reading to take * @param temp Reference in which to return the intake temperature - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a temp has been populated * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a unit or \a type is invalid or \a temp is NULL @@ -1777,8 +2919,8 @@ nvmlReturn_t DECLDIR nvmlUnitGetTemperature(nvmlUnit_t unit, unsigned int type, * * @param unit The identifier of the target unit * @param fanSpeeds Reference in which to return the fan speed information - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a fanSpeeds has been populated * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a unit is invalid or \a fanSpeeds is NULL @@ -1798,8 +2940,8 @@ nvmlReturn_t DECLDIR nvmlUnitGetFanSpeedInfo(nvmlUnit_t unit, nvmlUnitFanSpeeds_ * @param deviceCount Reference in which to provide the \a devices array size, and * to return the number of attached GPU devices * @param devices Reference in which to return the references to the attached GPU devices - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a deviceCount and \a devices have been populated * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a deviceCount indicates that the \a devices array is too small @@ -1808,39 +2950,21 @@ nvmlReturn_t DECLDIR nvmlUnitGetFanSpeedInfo(nvmlUnit_t unit, nvmlUnitFanSpeeds_ */ nvmlReturn_t DECLDIR nvmlUnitGetDevices(nvmlUnit_t unit, unsigned int *deviceCount, nvmlDevice_t *devices); -/** - * Retrieves the IDs and firmware versions for any Host Interface Cards (HICs) in the system. - * - * For S-class products. - * - * The \a hwbcCount argument is expected to be set to the size of the input \a hwbcEntries array. - * The HIC must be connected to an S-class system for it to be reported by this function. - * - * @param hwbcCount Size of hwbcEntries array - * @param hwbcEntries Array holding information about hwbc - * - * @return - * - \ref NVML_SUCCESS if \a hwbcCount and \a hwbcEntries have been populated - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if either \a hwbcCount or \a hwbcEntries is NULL - * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a hwbcCount indicates that the \a hwbcEntries array is too small - */ -nvmlReturn_t DECLDIR nvmlSystemGetHicVersion(unsigned int *hwbcCount, nvmlHwbcEntry_t *hwbcEntries); /** @} */ /***************************************************************************************************/ /** @defgroup nvmlDeviceQueries Device Queries * This chapter describes that queries that NVML can perform against each device. - * In each case the device is identified with an nvmlDevice_t handle. This handle is obtained by - * calling one of \ref nvmlDeviceGetHandleByIndex(), \ref nvmlDeviceGetHandleBySerial(), - * \ref nvmlDeviceGetHandleByPciBusId(). or \ref nvmlDeviceGetHandleByUUID(). + * In each case the device is identified with an nvmlDevice_t handle. This handle is obtained by + * calling one of \ref nvmlDeviceGetHandleByIndex_v2(), \ref nvmlDeviceGetHandleBySerial(), + * \ref nvmlDeviceGetHandleByPciBusId_v2(). or \ref nvmlDeviceGetHandleByUUID(). * @{ */ /***************************************************************************************************/ /** * Retrieves the number of compute devices in the system. A compute device is a single GPU. - * + * * For all products. * * Note: New nvmlDeviceGetCount_v2 (default in NVML 5.319) returns count of all devices in the system @@ -1851,34 +2975,54 @@ nvmlReturn_t DECLDIR nvmlSystemGetHicVersion(unsigned int *hwbcCount, nvmlHwbcEn * Old _v1 version of nvmlDeviceGetCount doesn't count devices that NVML has no permission to talk to. * * @param deviceCount Reference in which to return the number of accessible devices - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a deviceCount has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a deviceCount is NULL * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ -nvmlReturn_t DECLDIR nvmlDeviceGetCount(unsigned int *deviceCount); +nvmlReturn_t DECLDIR nvmlDeviceGetCount_v2(unsigned int *deviceCount); + +/** + * Get attributes (engine counts etc.) for the given NVML device handle. + * + * @note This API currently only supports MIG device handles. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device NVML device handle + * @param attributes Device attributes + * + * @return + * - \ref NVML_SUCCESS if \a device attributes were successfully retrieved + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device handle is invalid + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetAttributes_v2(nvmlDevice_t device, nvmlDeviceAttributes_t *attributes); /** * Acquire the handle for a particular device, based on its index. - * + * * For all products. * - * Valid indices are derived from the \a accessibleDevices count returned by - * \ref nvmlDeviceGetCount(). For example, if \a accessibleDevices is 2 the valid indices + * Valid indices are derived from the \a accessibleDevices count returned by + * \ref nvmlDeviceGetCount_v2(). For example, if \a accessibleDevices is 2 the valid indices * are 0 and 1, corresponding to GPU 0 and GPU 1. * * The order in which NVML enumerates devices has no guarantees of consistency between reboots. For that reason it - * is recommended that devices be looked up by their PCI ids or UUID. See - * \ref nvmlDeviceGetHandleByUUID() and \ref nvmlDeviceGetHandleByPciBusId(). + * is recommended that devices be looked up by their PCI ids or UUID. See + * \ref nvmlDeviceGetHandleByUUID() and \ref nvmlDeviceGetHandleByPciBusId_v2(). * * Note: The NVML index may not correlate with other APIs, such as the CUDA device index. * * Starting from NVML 5, this API causes NVML to initialize the target GPU * NVML may initialize additional GPUs if: * - The target GPU is an SLI slave - * + * * Note: New nvmlDeviceGetCount_v2 (default in NVML 5.319) returns count of all devices in the system * even if nvmlDeviceGetHandleByIndex_v2 returns NVML_ERROR_NO_PERMISSION for such device. * Update your code to handle this error, or use NVML 4.304 or older nvml header file. @@ -1892,8 +3036,8 @@ nvmlReturn_t DECLDIR nvmlDeviceGetCount(unsigned int *deviceCount); * * @param index The index of the target GPU, >= 0 and < \a accessibleDevices * @param device Reference in which to return the device handle - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a device has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a index is invalid or \a device is NULL @@ -1906,7 +3050,7 @@ nvmlReturn_t DECLDIR nvmlDeviceGetCount(unsigned int *deviceCount); * @see nvmlDeviceGetIndex * @see nvmlDeviceGetCount */ -nvmlReturn_t DECLDIR nvmlDeviceGetHandleByIndex(unsigned int index, nvmlDevice_t *device); +nvmlReturn_t DECLDIR nvmlDeviceGetHandleByIndex_v2(unsigned int index, nvmlDevice_t *device); /** * Acquire the handle for a particular device, based on its board serial number. @@ -1916,7 +3060,7 @@ nvmlReturn_t DECLDIR nvmlDeviceGetHandleByIndex(unsigned int index, nvmlDevice_t * This number corresponds to the value printed directly on the board, and to the value returned by * \ref nvmlDeviceGetSerial(). * - * @deprecated Since more than one GPU can exist on a single board this function is deprecated in favor + * @deprecated Since more than one GPU can exist on a single board this function is deprecated in favor * of \ref nvmlDeviceGetHandleByUUID. * For dual GPU boards this function will return NVML_ERROR_INVALID_ARGUMENT. * @@ -1925,8 +3069,8 @@ nvmlReturn_t DECLDIR nvmlDeviceGetHandleByIndex(unsigned int index, nvmlDevice_t * * @param serial The board serial number of the target GPU * @param device Reference in which to return the device handle - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a device has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a serial is invalid, \a device is NULL or more than one @@ -1947,13 +3091,13 @@ nvmlReturn_t DECLDIR nvmlDeviceGetHandleBySerial(const char *serial, nvmlDevice_ * * For all products. * - * @param uuid The UUID of the target GPU - * @param device Reference in which to return the device handle - * + * @param uuid The UUID of the target GPU or MIG instance + * @param device Reference in which to return the device handle or MIG device handle + * * Starting from NVML 5, this API causes NVML to initialize the target GPU * NVML may initialize additional GPUs as it searches for the target GPU * - * @return + * @return * - \ref NVML_SUCCESS if \a device has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a uuid is invalid or \a device is null @@ -1969,22 +3113,22 @@ nvmlReturn_t DECLDIR nvmlDeviceGetHandleByUUID(const char *uuid, nvmlDevice_t *d /** * Acquire the handle for a particular device, based on its PCI bus id. - * + * * For all products. * - * This value corresponds to the nvmlPciInfo_t::busId returned by \ref nvmlDeviceGetPciInfo(). + * This value corresponds to the nvmlPciInfo_t::busId returned by \ref nvmlDeviceGetPciInfo_v3(). * * Starting from NVML 5, this API causes NVML to initialize the target GPU * NVML may initialize additional GPUs if: * - The target GPU is an SLI slave * - * \note NVML 4.304 and older version of nvmlDeviceGetHandleByPciBusId"_v1" returns NVML_ERROR_NOT_FOUND + * \note NVML 4.304 and older version of nvmlDeviceGetHandleByPciBusId"_v1" returns NVML_ERROR_NOT_FOUND * instead of NVML_ERROR_NO_PERMISSION. * * @param pciBusId The PCI bus id of the target GPU * @param device Reference in which to return the device handle - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a device has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a pciBusId is invalid or \a device is NULL @@ -1995,22 +3139,25 @@ nvmlReturn_t DECLDIR nvmlDeviceGetHandleByUUID(const char *uuid, nvmlDevice_t *d * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ -nvmlReturn_t DECLDIR nvmlDeviceGetHandleByPciBusId(const char *pciBusId, nvmlDevice_t *device); +nvmlReturn_t DECLDIR nvmlDeviceGetHandleByPciBusId_v2(const char *pciBusId, nvmlDevice_t *device); /** - * Retrieves the name of this device. - * + * Retrieves the name of this device. + * * For all products. * * The name is an alphanumeric string that denotes a particular product, e.g. Tesla &tm; C2070. It will not - * exceed 64 characters in length (including the NULL terminator). See \ref - * nvmlConstants::NVML_DEVICE_NAME_BUFFER_SIZE. + * exceed 96 characters in length (including the NULL terminator). See \ref + * nvmlConstants::NVML_DEVICE_NAME_V2_BUFFER_SIZE. + * + * When used with MIG device handles the API returns MIG device names which can be used to identify devices + * based on their attributes. * * @param device The identifier of the target device * @param name Reference in which to return the product name * @param length The maximum allowed length of the string returned in \a name - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a name has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a name is NULL @@ -2043,21 +3190,25 @@ nvmlReturn_t DECLDIR nvmlDeviceGetBrand(nvmlDevice_t device, nvmlBrandType_t *ty * Retrieves the NVML index of this device. * * For all products. - * - * Valid indices are derived from the \a accessibleDevices count returned by - * \ref nvmlDeviceGetCount(). For example, if \a accessibleDevices is 2 the valid indices + * + * Valid indices are derived from the \a accessibleDevices count returned by + * \ref nvmlDeviceGetCount_v2(). For example, if \a accessibleDevices is 2 the valid indices * are 0 and 1, corresponding to GPU 0 and GPU 1. * * The order in which NVML enumerates devices has no guarantees of consistency between reboots. For that reason it - * is recommended that devices be looked up by their PCI ids or GPU UUID. See - * \ref nvmlDeviceGetHandleByPciBusId() and \ref nvmlDeviceGetHandleByUUID(). + * is recommended that devices be looked up by their PCI ids or GPU UUID. See + * \ref nvmlDeviceGetHandleByPciBusId_v2() and \ref nvmlDeviceGetHandleByUUID(). + * + * When used with MIG device handles this API returns indices that can be + * passed to \ref nvmlDeviceGetMigDeviceHandleByIndex to retrieve an identical handle. + * MIG device indices are unique within a device. * * Note: The NVML index may not correlate with other APIs, such as the CUDA device index. * * @param device The identifier of the target device * @param index Reference in which to return the NVML index of the device * - * @return + * @return * - \ref NVML_SUCCESS if \a index has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a index is NULL @@ -2081,8 +3232,8 @@ nvmlReturn_t DECLDIR nvmlDeviceGetIndex(nvmlDevice_t device, unsigned int *index * @param device The identifier of the target device * @param serial Reference in which to return the board/module serial number * @param length The maximum allowed length of the string returned in \a serial - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a serial has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a serial is NULL @@ -2093,20 +3244,129 @@ nvmlReturn_t DECLDIR nvmlDeviceGetIndex(nvmlDevice_t device, unsigned int *index */ nvmlReturn_t DECLDIR nvmlDeviceGetSerial(nvmlDevice_t device, char *serial, unsigned int length); +/* +* Get a unique identifier for the device module on the baseboard +* +* This API retrieves a unique identifier for each GPU module that exists on a given baseboard. +* For non-baseboard products, this ID would always be 0. +* +* @param device The identifier of the target device +* @param moduleId Unique identifier for the GPU module +* +* @return +* - \ref NVML_SUCCESS if \a moduleId has been successfully retrieved +* - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized +* - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a moduleId is invalid +* - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlDeviceGetModuleId(nvmlDevice_t device, unsigned int *moduleId); + /** - * Retrieves an array of unsigned ints (sized to cpuSetSize) of bitmasks with the ideal CPU affinity for the device + * Retrieves the Device's C2C Mode information + * + * @param device The identifier of the target device + * @param c2cModeInfo Output struct containing the device's C2C Mode info + * + * @return + * - \ref NVML_SUCCESS if \a C2C Mode Infor query is successful + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a serial is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetC2cModeInfoV(nvmlDevice_t device, nvmlC2cModeInfo_v1_t *c2cModeInfo); + +/***************************************************************************************************/ + +/** @defgroup nvmlAffinity CPU and Memory Affinity + * This chapter describes NVML operations that are associated with CPU and memory + * affinity. + * @{ + */ +/***************************************************************************************************/ + +//! Scope of NUMA node for affinity queries +#define NVML_AFFINITY_SCOPE_NODE 0 +//! Scope of processor socket for affinity queries +#define NVML_AFFINITY_SCOPE_SOCKET 1 + +typedef unsigned int nvmlAffinityScope_t; + +/** + * Retrieves an array of unsigned ints (sized to nodeSetSize) of bitmasks with + * the ideal memory affinity within node or socket for the device. + * For example, if NUMA node 0, 1 are ideal within the socket for the device and nodeSetSize == 1, + * result[0] = 0x3 + * + * \note If requested scope is not applicable to the target topology, the API + * will fall back to reporting the memory affinity for the immediate non-I/O + * ancestor of the device. + * + * For Kepler &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device The identifier of the target device + * @param nodeSetSize The size of the nodeSet array that is safe to access + * @param nodeSet Array reference in which to return a bitmask of NODEs, 64 NODEs per + * unsigned long on 64-bit machines, 32 on 32-bit machines + * @param scope Scope that change the default behavior + * + * @return + * - \ref NVML_SUCCESS if \a NUMA node Affinity has been filled + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, nodeSetSize == 0, nodeSet is NULL or scope is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ + +nvmlReturn_t DECLDIR nvmlDeviceGetMemoryAffinity(nvmlDevice_t device, unsigned int nodeSetSize, unsigned long *nodeSet, nvmlAffinityScope_t scope); + +/** + * Retrieves an array of unsigned ints (sized to cpuSetSize) of bitmasks with the + * ideal CPU affinity within node or socket for the device. * For example, if processors 0, 1, 32, and 33 are ideal for the device and cpuSetSize == 2, * result[0] = 0x3, result[1] = 0x3 * + * \note If requested scope is not applicable to the target topology, the API + * will fall back to reporting the CPU affinity for the immediate non-I/O + * ancestor of the device. + * * For Kepler &tm; or newer fully supported devices. * Supported on Linux only. * * @param device The identifier of the target device * @param cpuSetSize The size of the cpuSet array that is safe to access - * @param cpuSet Array reference in which to return a bitmask of CPUs, 64 CPUs per + * @param cpuSet Array reference in which to return a bitmask of CPUs, 64 CPUs per + * unsigned long on 64-bit machines, 32 on 32-bit machines + * @param scope Scope that change the default behavior + * + * @return + * - \ref NVML_SUCCESS if \a cpuAffinity has been filled + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, cpuSetSize == 0, cpuSet is NULL or sope is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ + +nvmlReturn_t DECLDIR nvmlDeviceGetCpuAffinityWithinScope(nvmlDevice_t device, unsigned int cpuSetSize, unsigned long *cpuSet, nvmlAffinityScope_t scope); + +/** + * Retrieves an array of unsigned ints (sized to cpuSetSize) of bitmasks with the ideal CPU affinity for the device + * For example, if processors 0, 1, 32, and 33 are ideal for the device and cpuSetSize == 2, + * result[0] = 0x3, result[1] = 0x3 + * This is equivalent to calling \ref nvmlDeviceGetCpuAffinityWithinScope with \ref NVML_AFFINITY_SCOPE_NODE. + * + * For Kepler &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device The identifier of the target device + * @param cpuSetSize The size of the cpuSet array that is safe to access + * @param cpuSet Array reference in which to return a bitmask of CPUs, 64 CPUs per * unsigned long on 64-bit machines, 32 on 32-bit machines * - * @return + * @return * - \ref NVML_SUCCESS if \a cpuAffinity has been filled * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, cpuSetSize == 0, or cpuSet is NULL @@ -2117,17 +3377,17 @@ nvmlReturn_t DECLDIR nvmlDeviceGetSerial(nvmlDevice_t device, char *serial, unsi nvmlReturn_t DECLDIR nvmlDeviceGetCpuAffinity(nvmlDevice_t device, unsigned int cpuSetSize, unsigned long *cpuSet); /** - * Sets the ideal affinity for the calling thread and device using the guidelines - * given in nvmlDeviceGetCpuAffinity(). Note, this is a change as of version 8.0. + * Sets the ideal affinity for the calling thread and device using the guidelines + * given in nvmlDeviceGetCpuAffinity(). Note, this is a change as of version 8.0. * Older versions set the affinity for a calling process and all children. - * Currently supports up to 64 processors. + * Currently supports up to 1024 processors. * * For Kepler &tm; or newer fully supported devices. * Supported on Linux only. * * @param device The identifier of the target device * - * @return + * @return * - \ref NVML_SUCCESS if the calling process has been successfully bound * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid @@ -2146,7 +3406,7 @@ nvmlReturn_t DECLDIR nvmlDeviceSetCpuAffinity(nvmlDevice_t device); * * @param device The identifier of the target device * - * @return + * @return * - \ref NVML_SUCCESS if the calling process has been successfully unbound * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized @@ -2169,6 +3429,8 @@ nvmlReturn_t DECLDIR nvmlDeviceClearCpuAffinity(nvmlDevice_t device); * - \ref NVML_ERROR_NOT_SUPPORTED if the device or OS does not support this feature * - \ref NVML_ERROR_UNKNOWN an error has occurred in underlying topology discovery */ + +/** @} */ nvmlReturn_t DECLDIR nvmlDeviceGetTopologyCommonAncestor(nvmlDevice_t device1, nvmlDevice_t device2, nvmlGpuTopologyLevel_t *pathInfo); /** @@ -2178,7 +3440,7 @@ nvmlReturn_t DECLDIR nvmlDeviceGetTopologyCommonAncestor(nvmlDevice_t device1, n * * @param device The identifier of the first device * @param level The \ref nvmlGpuTopologyLevel_t level to search for other GPUs - * @param count When zero, is set to the number of matching GPUs such that \a deviceArray + * @param count When zero, is set to the number of matching GPUs such that \a deviceArray * can be malloc'd. When non-zero, \a deviceArray will be filled with \a count * number of device handles. * @param deviceArray An array of device handles for GPUs found at \a level @@ -2192,37 +3454,18 @@ nvmlReturn_t DECLDIR nvmlDeviceGetTopologyCommonAncestor(nvmlDevice_t device1, n nvmlReturn_t DECLDIR nvmlDeviceGetTopologyNearestGpus(nvmlDevice_t device, nvmlGpuTopologyLevel_t level, unsigned int *count, nvmlDevice_t *deviceArray); /** - * Retrieve the set of GPUs that have a CPU affinity with the given CPU number - * For all products. - * Supported on Linux only. + * Retrieve the status for a given p2p capability index between a given pair of GPU * - * @param cpuNumber The CPU number - * @param count When zero, is set to the number of matching GPUs such that \a deviceArray - * can be malloc'd. When non-zero, \a deviceArray will be filled with \a count - * number of device handles. - * @param deviceArray An array of device handles for GPUs found with affinity to \a cpuNumber - * - * @return - * - \ref NVML_SUCCESS if \a deviceArray or \a count (if initially zero) has been set - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a cpuNumber, or \a count is invalid, or \a deviceArray is NULL with a non-zero \a count - * - \ref NVML_ERROR_NOT_SUPPORTED if the device or OS does not support this feature - * - \ref NVML_ERROR_UNKNOWN an error has occurred in underlying topology discovery - */ -nvmlReturn_t DECLDIR nvmlSystemGetTopologyGpuSet(unsigned int cpuNumber, unsigned int *count, nvmlDevice_t *deviceArray); - -/** - * Retrieve the status for a given p2p capability index between a given pair of GPU - * - * @param device1 The first device + * @param device1 The first device * @param device2 The second device * @param p2pIndex p2p Capability Index being looked for between \a device1 and \a device2 - * @param p2pStatus Reference in which to return the status of the \a p2pIndex + * @param p2pStatus Reference in which to return the status of the \a p2pIndex * between \a device1 and \a device2 - * @return + * @return * - \ref NVML_SUCCESS if \a p2pStatus has been populated * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device1 or \a device2 or \a p2pIndex is invalid or \a p2pStatus is NULL * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ + */ nvmlReturn_t DECLDIR nvmlDeviceGetP2PStatus(nvmlDevice_t device1, nvmlDevice_t device2, nvmlGpuP2PCapsIndex_t p2pIndex,nvmlGpuP2PStatus_t *p2pStatus); /** @@ -2232,18 +3475,21 @@ nvmlReturn_t DECLDIR nvmlDeviceGetP2PStatus(nvmlDevice_t device1, nvmlDevice_t d * For all products. * * The UUID is a globally unique identifier. It is the only available identifier for pre-Fermi-architecture products. - * It does NOT correspond to any identifier printed on the board. It will not exceed 80 characters in length - * (including the NULL terminator). See \ref nvmlConstants::NVML_DEVICE_UUID_BUFFER_SIZE. + * It does NOT correspond to any identifier printed on the board. It will not exceed 96 characters in length + * (including the NULL terminator). See \ref nvmlConstants::NVML_DEVICE_UUID_V2_BUFFER_SIZE. + * + * When used with MIG device handles the API returns globally unique UUIDs which can be used to identify MIG + * devices across both GPU and MIG devices. UUIDs are immutable for the lifetime of a MIG device. * * @param device The identifier of the target device * @param uuid Reference in which to return the GPU UUID * @param length The maximum allowed length of the string returned in \a uuid - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a uuid has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a uuid is NULL - * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible * - \ref NVML_ERROR_UNKNOWN on any unexpected error @@ -2251,7 +3497,7 @@ nvmlReturn_t DECLDIR nvmlDeviceGetP2PStatus(nvmlDevice_t device1, nvmlDevice_t d nvmlReturn_t DECLDIR nvmlDeviceGetUUID(nvmlDevice_t device, char *uuid, unsigned int length); /** - * Retrieves minor number for the device. The minor number for the device is such that the Nvidia device node file for + * Retrieves minor number for the device. The minor number for the device is such that the Nvidia device node file for * each GPU will have the form /dev/nvidia[minor number]. * * For all products. @@ -2293,7 +3539,7 @@ nvmlReturn_t DECLDIR nvmlDeviceGetBoardPartNumber(nvmlDevice_t device, char* par * * For all products with an inforom. * - * Fermi and higher parts have non-volatile on-board memory for persisting device info, such as aggregate + * Fermi and higher parts have non-volatile on-board memory for persisting device info, such as aggregate * ECC counts. The version of the data structures in this memory may change from time to time. It will not * exceed 16 characters in length (including the NULL terminator). * See \ref nvmlConstants::NVML_DEVICE_INFOROM_VERSION_BUFFER_SIZE. @@ -2305,11 +3551,11 @@ nvmlReturn_t DECLDIR nvmlDeviceGetBoardPartNumber(nvmlDevice_t device, char* par * @param version Reference in which to return the infoROM version * @param length The maximum allowed length of the string returned in \a version * - * @return + * @return * - \ref NVML_SUCCESS if \a version has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a version is NULL - * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have an infoROM * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible * - \ref NVML_ERROR_UNKNOWN on any unexpected error @@ -2323,7 +3569,7 @@ nvmlReturn_t DECLDIR nvmlDeviceGetInforomVersion(nvmlDevice_t device, nvmlInforo * * For all products with an inforom. * - * Image version just like VBIOS version uniquely describes the exact version of the infoROM flashed on the board + * Image version just like VBIOS version uniquely describes the exact version of the infoROM flashed on the board * in contrast to infoROM object version which is only an indicator of supported features. * Version string will not exceed 16 characters in length (including the NULL terminator). * See \ref nvmlConstants::NVML_DEVICE_INFOROM_VERSION_BUFFER_SIZE. @@ -2332,11 +3578,11 @@ nvmlReturn_t DECLDIR nvmlDeviceGetInforomVersion(nvmlDevice_t device, nvmlInforo * @param version Reference in which to return the infoROM image version * @param length The maximum allowed length of the string returned in \a version * - * @return + * @return * - \ref NVML_SUCCESS if \a version has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a version is NULL - * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have an infoROM * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible * - \ref NVML_ERROR_UNKNOWN on any unexpected error @@ -2357,14 +3603,14 @@ nvmlReturn_t DECLDIR nvmlDeviceGetInforomImageVersion(nvmlDevice_t device, char * @param device The identifier of the target device * @param checksum Reference in which to return the infoROM configuration checksum * - * @return + * @return * - \ref NVML_SUCCESS if \a checksum has been set * - \ref NVML_ERROR_CORRUPTED_INFOROM if the device's checksum couldn't be retrieved due to infoROM corruption * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a checksum is NULL * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ nvmlReturn_t DECLDIR nvmlDeviceGetInforomConfigurationChecksum(nvmlDevice_t device, unsigned int *checksum); @@ -2375,16 +3621,37 @@ nvmlReturn_t DECLDIR nvmlDeviceGetInforomConfigurationChecksum(nvmlDevice_t devi * * @param device The identifier of the target device * - * @return + * @return * - \ref NVML_SUCCESS if infoROM is not corrupted * - \ref NVML_ERROR_CORRUPTED_INFOROM if the device's infoROM is corrupted * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ nvmlReturn_t DECLDIR nvmlDeviceValidateInforom(nvmlDevice_t device); +/** + * Retrieves the timestamp and the duration of the last flush of the BBX (blackbox) infoROM object during the current run. + * + * For all products with an inforom. + * + * @param device The identifier of the target device + * @param timestamp The start timestamp of the last BBX Flush + * @param durationUs The duration (us) of the last BBX Flush + * + * @return + * - \ref NVML_SUCCESS if \a timestamp and \a durationUs are successfully retrieved + * - \ref NVML_ERROR_NOT_READY if the BBX object has not been flushed yet + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have an infoROM + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetInforomVersion + */ +nvmlReturn_t DECLDIR nvmlDeviceGetLastBBXFlushTime(nvmlDevice_t device, unsigned long long *timestamp, + unsigned long *durationUs); + /** * Retrieves the display mode for the device. * @@ -2397,8 +3664,8 @@ nvmlReturn_t DECLDIR nvmlDeviceValidateInforom(nvmlDevice_t device); * * @param device The identifier of the target device * @param display Reference in which to return the display mode - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a display has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a display is NULL @@ -2422,8 +3689,8 @@ nvmlReturn_t DECLDIR nvmlDeviceGetDisplayMode(nvmlDevice_t device, nvmlEnableSta * * @param device The identifier of the target device * @param isActive Reference in which to return the display active state - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a isActive has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a isActive is NULL @@ -2439,15 +3706,15 @@ nvmlReturn_t DECLDIR nvmlDeviceGetDisplayActive(nvmlDevice_t device, nvmlEnableS * For all products. * For Linux only. * - * When driver persistence mode is enabled the driver software state is not torn down when the last - * client disconnects. By default this feature is disabled. + * When driver persistence mode is enabled the driver software state is not torn down when the last + * client disconnects. By default this feature is disabled. * * See \ref nvmlEnableState_t for details on allowed modes. * * @param device The identifier of the target device * @param mode Reference in which to return the current driver persistence mode - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a mode has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode is NULL @@ -2461,35 +3728,35 @@ nvmlReturn_t DECLDIR nvmlDeviceGetPersistenceMode(nvmlDevice_t device, nvmlEnabl /** * Retrieves the PCI attributes of this device. - * + * * For all products. * * See \ref nvmlPciInfo_t for details on the available PCI info. * * @param device The identifier of the target device * @param pci Reference in which to return the PCI info - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a pci has been populated * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pci is NULL * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ -nvmlReturn_t DECLDIR nvmlDeviceGetPciInfo(nvmlDevice_t device, nvmlPciInfo_t *pci); +nvmlReturn_t DECLDIR nvmlDeviceGetPciInfo_v3(nvmlDevice_t device, nvmlPciInfo_t *pci); /** * Retrieves the maximum PCIe link generation possible with this device and system * * I.E. for a generation 2 PCIe device attached to a generation 1 PCIe bus the max link generation this function will * report is generation 1. - * + * * For Fermi &tm; or newer fully supported devices. - * + * * @param device The identifier of the target device * @param maxLinkGen Reference in which to return the max PCIe link generation - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a maxLinkGen has been populated * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a maxLinkGen is null @@ -2499,18 +3766,36 @@ nvmlReturn_t DECLDIR nvmlDeviceGetPciInfo(nvmlDevice_t device, nvmlPciInfo_t *pc */ nvmlReturn_t DECLDIR nvmlDeviceGetMaxPcieLinkGeneration(nvmlDevice_t device, unsigned int *maxLinkGen); +/** + * Retrieves the maximum PCIe link generation supported by this device + * + * For Fermi &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param maxLinkGenDevice Reference in which to return the max PCIe link generation + * + * @return + * - \ref NVML_SUCCESS if \a maxLinkGenDevice has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a maxLinkGenDevice is null + * - \ref NVML_ERROR_NOT_SUPPORTED if PCIe link information is not available + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuMaxPcieLinkGeneration(nvmlDevice_t device, unsigned int *maxLinkGenDevice); + /** * Retrieves the maximum PCIe link width possible with this device and system * * I.E. for a device with a 16x PCIe bus width attached to a 8x PCIe system bus this function will report * a max link width of 8. - * + * * For Fermi &tm; or newer fully supported devices. - * + * * @param device The identifier of the target device * @param maxLinkWidth Reference in which to return the max PCIe link generation - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a maxLinkWidth has been populated * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a maxLinkWidth is null @@ -2522,13 +3807,13 @@ nvmlReturn_t DECLDIR nvmlDeviceGetMaxPcieLinkWidth(nvmlDevice_t device, unsigned /** * Retrieves the current PCIe link generation - * + * * For Fermi &tm; or newer fully supported devices. - * + * * @param device The identifier of the target device * @param currLinkGen Reference in which to return the current PCIe link generation - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a currLinkGen has been populated * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a currLinkGen is null @@ -2540,13 +3825,13 @@ nvmlReturn_t DECLDIR nvmlDeviceGetCurrPcieLinkGeneration(nvmlDevice_t device, un /** * Retrieves the current PCIe link width - * + * * For Fermi &tm; or newer fully supported devices. - * + * * @param device The identifier of the target device * @param currLinkWidth Reference in which to return the current PCIe link generation - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a currLinkWidth has been populated * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a currLinkWidth is null @@ -2558,7 +3843,7 @@ nvmlReturn_t DECLDIR nvmlDeviceGetCurrPcieLinkWidth(nvmlDevice_t device, unsigne /** * Retrieve PCIe utilization information. - * This function is querying a byte counter over a 20ms interval and thus is the + * This function is querying a byte counter over a 20ms interval and thus is the * PCIe throughput over that interval. * * For Maxwell &tm; or newer fully supported devices. @@ -2579,7 +3864,7 @@ nvmlReturn_t DECLDIR nvmlDeviceGetCurrPcieLinkWidth(nvmlDevice_t device, unsigne */ nvmlReturn_t DECLDIR nvmlDeviceGetPcieThroughput(nvmlDevice_t device, nvmlPcieUtilCounter_t counter, unsigned int *value); -/** +/** * Retrieve the PCIe replay counter. * * For Kepler &tm; or newer fully supported devices. @@ -2588,9 +3873,9 @@ nvmlReturn_t DECLDIR nvmlDeviceGetPcieThroughput(nvmlDevice_t device, nvmlPcieUt * @param value Reference in which to return the counter's value * * @return - * - \ref NVML_SUCCESS if \a value and \a rollover have been set + * - \ref NVML_SUCCESS if \a value has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a value or \a rollover are NULL + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a value is NULL * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible * - \ref NVML_ERROR_UNKNOWN on any unexpected error @@ -2607,8 +3892,8 @@ nvmlReturn_t DECLDIR nvmlDeviceGetPcieReplayCounter(nvmlDevice_t device, unsigne * @param device The identifier of the target device * @param type Identify which clock domain to query * @param clock Reference in which to return the clock speed in MHz - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a clock has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a clock is NULL @@ -2631,8 +3916,8 @@ nvmlReturn_t DECLDIR nvmlDeviceGetClockInfo(nvmlDevice_t device, nvmlClockType_t * @param device The identifier of the target device * @param type Identify which clock domain to query * @param clock Reference in which to return the clock speed in MHz - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a clock has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a clock is NULL @@ -2642,6 +3927,20 @@ nvmlReturn_t DECLDIR nvmlDeviceGetClockInfo(nvmlDevice_t device, nvmlClockType_t */ nvmlReturn_t DECLDIR nvmlDeviceGetMaxClockInfo(nvmlDevice_t device, nvmlClockType_t type, unsigned int *clock); +/** + * Retrieve the GPCCLK VF offset value + * @param[in] device The identifier of the target device + * @param[out] offset The retrieved GPCCLK VF offset value + * + * @return + * - \ref NVML_SUCCESS if \a offset has been successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a offset is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpcClkVfOffset(nvmlDevice_t device, int *offset); + /** * Retrieves the current setting of a clock that applications will use unless an overspec situation occurs. * Can be changed using \ref nvmlDeviceSetApplicationsClocks. @@ -2651,8 +3950,8 @@ nvmlReturn_t DECLDIR nvmlDeviceGetMaxClockInfo(nvmlDevice_t device, nvmlClockTyp * @param device The identifier of the target device * @param clockType Identify which clock domain to query * @param clockMHz Reference in which to return the clock in MHz - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a clockMHz has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a clockMHz is NULL or \a clockType is invalid @@ -2663,7 +3962,7 @@ nvmlReturn_t DECLDIR nvmlDeviceGetMaxClockInfo(nvmlDevice_t device, nvmlClockTyp nvmlReturn_t DECLDIR nvmlDeviceGetApplicationsClock(nvmlDevice_t device, nvmlClockType_t clockType, unsigned int *clockMHz); /** - * Retrieves the default applications clock that GPU boots with or + * Retrieves the default applications clock that GPU boots with or * defaults to after \ref nvmlDeviceResetApplicationsClocks call. * * For Kepler &tm; or newer fully supported devices. @@ -2671,8 +3970,8 @@ nvmlReturn_t DECLDIR nvmlDeviceGetApplicationsClock(nvmlDevice_t device, nvmlClo * @param device The identifier of the target device * @param clockType Identify which clock domain to query * @param clockMHz Reference in which to return the default clock in MHz - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a clockMHz has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a clockMHz is NULL or \a clockType is invalid @@ -2684,33 +3983,6 @@ nvmlReturn_t DECLDIR nvmlDeviceGetApplicationsClock(nvmlDevice_t device, nvmlClo */ nvmlReturn_t DECLDIR nvmlDeviceGetDefaultApplicationsClock(nvmlDevice_t device, nvmlClockType_t clockType, unsigned int *clockMHz); -/** - * Resets the application clock to the default value - * - * This is the applications clock that will be used after system reboot or driver reload. - * Default value is constant, but the current value an be changed using \ref nvmlDeviceSetApplicationsClocks. - * - * On Pascal and newer hardware, if clocks were previously locked with \ref nvmlDeviceSetApplicationsClocks, - * this call will unlock clocks. This returns clocks their default behavior ofautomatically boosting above - * base clocks as thermal limits allow. - * - * @see nvmlDeviceGetApplicationsClock - * @see nvmlDeviceSetApplicationsClocks - * - * For Fermi &tm; or newer non-GeForce fully supported devices and Maxwell or newer GeForce devices. - * - * @param device The identifier of the target device - * - * @return - * - \ref NVML_SUCCESS if new settings were successfully set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceResetApplicationsClocks(nvmlDevice_t device); - /** * Retrieves the clock speed for the clock specified by the clock type and clock ID. * @@ -2759,9 +4031,9 @@ nvmlReturn_t DECLDIR nvmlDeviceGetMaxCustomerBoostClock(nvmlDevice_t device, nvm * @param count Reference in which to provide the \a clocksMHz array size, and * to return the number of elements * @param clocksMHz Reference in which to return the clock in MHz - * - * @return - * - \ref NVML_SUCCESS if \a count and \a clocksMHz have been populated + * + * @return + * - \ref NVML_SUCCESS if \a count and \a clocksMHz have been populated * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a count is NULL * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature @@ -2785,14 +4057,14 @@ nvmlReturn_t DECLDIR nvmlDeviceGetSupportedMemoryClocks(nvmlDevice_t device, uns * @param count Reference in which to provide the \a clocksMHz array size, and * to return the number of elements * @param clocksMHz Reference in which to return the clocks in MHz - * - * @return - * - \ref NVML_SUCCESS if \a count and \a clocksMHz have been populated + * + * @return + * - \ref NVML_SUCCESS if \a count and \a clocksMHz have been populated * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_NOT_FOUND if the specified \a memoryClockMHz is not a supported frequency * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a clock is NULL * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a count is too small + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a count is too small * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible * - \ref NVML_ERROR_UNKNOWN on any unexpected error * @@ -2829,6 +4101,2568 @@ nvmlReturn_t DECLDIR nvmlDeviceGetSupportedGraphicsClocks(nvmlDevice_t device, u */ nvmlReturn_t DECLDIR nvmlDeviceGetAutoBoostedClocksEnabled(nvmlDevice_t device, nvmlEnableState_t *isEnabled, nvmlEnableState_t *defaultIsEnabled); +/** + * Retrieves the intended operating speed of the device's fan. + * + * Note: The reported speed is the intended fan speed. If the fan is physically blocked and unable to spin, the + * output will not match the actual fan speed. + * + * For all discrete products with dedicated fans. + * + * The fan speed is expressed as a percentage of the product's maximum noise tolerance fan speed. + * This value may exceed 100% in certain cases. + * + * @param device The identifier of the target device + * @param speed Reference in which to return the fan speed percentage + * + * @return + * - \ref NVML_SUCCESS if \a speed has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a speed is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have a fan + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetFanSpeed(nvmlDevice_t device, unsigned int *speed); + + +/** + * Retrieves the intended operating speed of the device's specified fan. + * + * Note: The reported speed is the intended fan speed. If the fan is physically blocked and unable to spin, the + * output will not match the actual fan speed. + * + * For all discrete products with dedicated fans. + * + * The fan speed is expressed as a percentage of the product's maximum noise tolerance fan speed. + * This value may exceed 100% in certain cases. + * + * @param device The identifier of the target device + * @param fan The index of the target fan, zero indexed. + * @param speed Reference in which to return the fan speed percentage + * + * @return + * - \ref NVML_SUCCESS if \a speed has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a fan is not an acceptable index, or \a speed is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have a fan or is newer than Maxwell + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetFanSpeed_v2(nvmlDevice_t device, unsigned int fan, unsigned int * speed); + +/** + * Retrieves the intended target speed of the device's specified fan. + * + * Normally, the driver dynamically adjusts the fan based on + * the needs of the GPU. But when user set fan speed using nvmlDeviceSetFanSpeed_v2, + * the driver will attempt to make the fan achieve the setting in + * nvmlDeviceSetFanSpeed_v2. The actual current speed of the fan + * is reported in nvmlDeviceGetFanSpeed_v2. + * + * For all discrete products with dedicated fans. + * + * The fan speed is expressed as a percentage of the product's maximum noise tolerance fan speed. + * This value may exceed 100% in certain cases. + * + * @param device The identifier of the target device + * @param fan The index of the target fan, zero indexed. + * @param targetSpeed Reference in which to return the fan speed percentage + * + * @return + * - \ref NVML_SUCCESS if \a speed has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a fan is not an acceptable index, or \a speed is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have a fan or is newer than Maxwell + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetTargetFanSpeed(nvmlDevice_t device, unsigned int fan, unsigned int *targetSpeed); + +/** + * Retrieves the min and max fan speed that user can set for the GPU fan. + * + * For all cuda-capable discrete products with fans + * + * @param device The identifier of the target device + * @param minSpeed The minimum speed allowed to set + * @param maxSpeed The maximum speed allowed to set + * + * return + * NVML_SUCCESS if speed has been adjusted + * NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * NVML_ERROR_INVALID_ARGUMENT if device is invalid + * NVML_ERROR_NOT_SUPPORTED if the device does not support this + * (doesn't have fans) + * NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMinMaxFanSpeed(nvmlDevice_t device, unsigned int * minSpeed, + unsigned int * maxSpeed); + +/** + * Gets current fan control policy. + * + * For Maxwell &tm; or newer fully supported devices. + * + * For all cuda-capable discrete products with fans + * + * device The identifier of the target \a device + * policy Reference in which to return the fan control \a policy + * + * return + * NVML_SUCCESS if \a policy has been populated + * NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a policy is null or the \a fan given doesn't reference + * a fan that exists. + * NVML_ERROR_NOT_SUPPORTED if the \a device is older than Maxwell + * NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetFanControlPolicy_v2(nvmlDevice_t device, unsigned int fan, + nvmlFanControlPolicy_t *policy); + +/** + * Retrieves the number of fans on the device. + * + * For all discrete products with dedicated fans. + * + * @param device The identifier of the target device + * @param numFans The number of fans + * + * @return + * - \ref NVML_SUCCESS if \a fan number query was successful + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a numFans is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have a fan + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetNumFans(nvmlDevice_t device, unsigned int *numFans); + +/** + * Retrieves the current temperature readings for the device, in degrees C. + * + * For all products. + * + * See \ref nvmlTemperatureSensors_t for details on available temperature sensors. + * + * @param device The identifier of the target device + * @param sensorType Flag that indicates which sensor reading to retrieve + * @param temp Reference in which to return the temperature reading + * + * @return + * - \ref NVML_SUCCESS if \a temp has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a sensorType is invalid or \a temp is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have the specified sensor + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetTemperature(nvmlDevice_t device, nvmlTemperatureSensors_t sensorType, unsigned int *temp); + +/** + * Retrieves the temperature threshold for the GPU with the specified threshold type in degrees C. + * + * For Kepler &tm; or newer fully supported devices. + * + * See \ref nvmlTemperatureThresholds_t for details on available temperature thresholds. + * + * Note: This API is no longer the preferred interface for retrieving the following temperature thresholds + * on Ada and later architectures: NVML_TEMPERATURE_THRESHOLD_SHUTDOWN, NVML_TEMPERATURE_THRESHOLD_SLOWDOWN, + * NVML_TEMPERATURE_THRESHOLD_MEM_MAX and NVML_TEMPERATURE_THRESHOLD_GPU_MAX. + * + * Support for reading these temperature thresholds for Ada and later architectures would be removed from this + * API in future releases. Please use \ref nvmlDeviceGetFieldValues with NVML_FI_DEV_TEMPERATURE_* fields to retrieve + * temperature thresholds on these architectures. + * + * @param device The identifier of the target device + * @param thresholdType The type of threshold value queried + * @param temp Reference in which to return the temperature reading + * @return + * - \ref NVML_SUCCESS if \a temp has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a thresholdType is invalid or \a temp is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have a temperature sensor or is unsupported + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetTemperatureThreshold(nvmlDevice_t device, nvmlTemperatureThresholds_t thresholdType, unsigned int *temp); + +/** + * Used to execute a list of thermal system instructions. + * + * @param device The identifier of the target device + * @param sensorIndex The index of the thermal sensor + * @param pThermalSettings Reference in which to return the thermal sensor information + * + * @return + * - \ref NVML_SUCCESS if \a pThermalSettings has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pThermalSettings is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetThermalSettings(nvmlDevice_t device, unsigned int sensorIndex, nvmlGpuThermalSettings_t *pThermalSettings); + +/** + * Retrieves the current performance state for the device. + * + * For Fermi &tm; or newer fully supported devices. + * + * See \ref nvmlPstates_t for details on allowed performance states. + * + * @param device The identifier of the target device + * @param pState Reference in which to return the performance state reading + * + * @return + * - \ref NVML_SUCCESS if \a pState has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pState is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPerformanceState(nvmlDevice_t device, nvmlPstates_t *pState); + +/** + * Retrieves current clocks event reasons. + * + * For all fully supported products. + * + * \note More than one bit can be enabled at the same time. Multiple reasons can be affecting clocks at once. + * + * @param device The identifier of the target device + * @param clocksEventReasons Reference in which to return bitmask of active clocks event + * reasons + * + * @return + * - \ref NVML_SUCCESS if \a clocksEventReasons has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a clocksEventReasons is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlClocksEventReasons + * @see nvmlDeviceGetSupportedClocksEventReasons + */ +nvmlReturn_t DECLDIR nvmlDeviceGetCurrentClocksEventReasons(nvmlDevice_t device, unsigned long long *clocksEventReasons); + +/** + * @deprecated Use \ref nvmlDeviceGetCurrentClocksEventReasons instead + */ +nvmlReturn_t DECLDIR nvmlDeviceGetCurrentClocksThrottleReasons(nvmlDevice_t device, unsigned long long *clocksThrottleReasons); + +/** + * Retrieves bitmask of supported clocks event reasons that can be returned by + * \ref nvmlDeviceGetCurrentClocksEventReasons + * + * For all fully supported products. + * + * This method is not supported in virtual machines running virtual GPU (vGPU). + * + * @param device The identifier of the target device + * @param supportedClocksEventReasons Reference in which to return bitmask of supported + * clocks event reasons + * + * @return + * - \ref NVML_SUCCESS if \a supportedClocksEventReasons has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a supportedClocksEventReasons is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlClocksEventReasons + * @see nvmlDeviceGetCurrentClocksEventReasons + */ +nvmlReturn_t DECLDIR nvmlDeviceGetSupportedClocksEventReasons(nvmlDevice_t device, unsigned long long *supportedClocksEventReasons); + +/** + * @deprecated Use \ref nvmlDeviceGetSupportedClocksEventReasons instead + */ +nvmlReturn_t DECLDIR nvmlDeviceGetSupportedClocksThrottleReasons(nvmlDevice_t device, unsigned long long *supportedClocksThrottleReasons); + +/** + * Deprecated: Use \ref nvmlDeviceGetPerformanceState. This function exposes an incorrect generalization. + * + * Retrieve the current performance state for the device. + * + * For Fermi &tm; or newer fully supported devices. + * + * See \ref nvmlPstates_t for details on allowed performance states. + * + * @param device The identifier of the target device + * @param pState Reference in which to return the performance state reading + * + * @return + * - \ref NVML_SUCCESS if \a pState has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pState is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPowerState(nvmlDevice_t device, nvmlPstates_t *pState); + +/** + * Retrieve performance monitor samples from the associated subdevice. + * + * @param device + * @param pDynamicPstatesInfo + * + * @return + * - \ref NVML_SUCCESS if \a pDynamicPstatesInfo has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pDynamicPstatesInfo is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetDynamicPstatesInfo(nvmlDevice_t device, nvmlGpuDynamicPstatesInfo_t *pDynamicPstatesInfo); + +/** + * Retrieve the MemClk (Memory Clock) VF offset value. + * @param[in] device The identifier of the target device + * @param[out] offset The retrieved MemClk VF offset value + * + * @return + * - \ref NVML_SUCCESS if \a offset has been successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a offset is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMemClkVfOffset(nvmlDevice_t device, int *offset); + +/** + * Retrieve min and max clocks of some clock domain for a given PState + * + * @param device The identifier of the target device + * @param type Clock domain + * @param pstate PState to query + * @param minClockMHz Reference in which to return min clock frequency + * @param maxClockMHz Reference in which to return max clock frequency + * + * @return + * - \ref NVML_SUCCESS if everything worked + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a type or \a pstate are invalid or both + * \a minClockMHz and \a maxClockMHz are NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMinMaxClockOfPState(nvmlDevice_t device, nvmlClockType_t type, nvmlPstates_t pstate, + unsigned int * minClockMHz, unsigned int * maxClockMHz); + +/** + * Get all supported Performance States (P-States) for the device. + * + * The returned array would contain a contiguous list of valid P-States supported by + * the device. If the number of supported P-States is fewer than the size of the array + * supplied missing elements would contain \a NVML_PSTATE_UNKNOWN. + * + * The number of elements in the returned list will never exceed \a NVML_MAX_GPU_PERF_PSTATES. + * + * @param device The identifier of the target device + * @param pstates Container to return the list of performance states + * supported by device + * @param size Size of the supplied \a pstates array in bytes + * + * @return + * - \ref NVML_SUCCESS if \a pstates array has been retrieved + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if the the container supplied was not large enough to + * hold the resulting list + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a pstates is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support performance state readings + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetSupportedPerformanceStates(nvmlDevice_t device, + nvmlPstates_t *pstates, unsigned int size); + +/** + * Retrieve the GPCCLK min max VF offset value. + * @param[in] device The identifier of the target device + * @param[out] minOffset The retrieved GPCCLK VF min offset value + * @param[out] maxOffset The retrieved GPCCLK VF max offset value + * + * @return + * - \ref NVML_SUCCESS if \a offset has been successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a offset is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpcClkMinMaxVfOffset(nvmlDevice_t device, + int *minOffset, int *maxOffset); + +/** + * Retrieve the MemClk (Memory Clock) min max VF offset value. + * @param[in] device The identifier of the target device + * @param[out] minOffset The retrieved MemClk VF min offset value + * @param[out] maxOffset The retrieved MemClk VF max offset value + * + * @return + * - \ref NVML_SUCCESS if \a offset has been successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a offset is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMemClkMinMaxVfOffset(nvmlDevice_t device, + int *minOffset, int *maxOffset); + +/** + * This API has been deprecated. + * + * Retrieves the power management mode associated with this device. + * + * For products from the Fermi family. + * - Requires \a NVML_INFOROM_POWER version 3.0 or higher. + * + * For from the Kepler or newer families. + * - Does not require \a NVML_INFOROM_POWER object. + * + * This flag indicates whether any power management algorithm is currently active on the device. An + * enabled state does not necessarily mean the device is being actively throttled -- only that + * that the driver will do so if the appropriate conditions are met. + * + * See \ref nvmlEnableState_t for details on allowed modes. + * + * @param device The identifier of the target device + * @param mode Reference in which to return the current power management mode + * + * @return + * - \ref NVML_SUCCESS if \a mode has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPowerManagementMode(nvmlDevice_t device, nvmlEnableState_t *mode); + +/** + * Retrieves the power management limit associated with this device. + * + * For Fermi &tm; or newer fully supported devices. + * + * The power limit defines the upper boundary for the card's power draw. If + * the card's total power draw reaches this limit the power management algorithm kicks in. + * + * This reading is only available if power management mode is supported. + * See \ref nvmlDeviceGetPowerManagementMode. + * + * @param device The identifier of the target device + * @param limit Reference in which to return the power management limit in milliwatts + * + * @return + * - \ref NVML_SUCCESS if \a limit has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a limit is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPowerManagementLimit(nvmlDevice_t device, unsigned int *limit); + +/** + * Retrieves information about possible values of power management limits on this device. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param minLimit Reference in which to return the minimum power management limit in milliwatts + * @param maxLimit Reference in which to return the maximum power management limit in milliwatts + * + * @return + * - \ref NVML_SUCCESS if \a minLimit and \a maxLimit have been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a minLimit or \a maxLimit is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceSetPowerManagementLimit + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPowerManagementLimitConstraints(nvmlDevice_t device, unsigned int *minLimit, unsigned int *maxLimit); + +/** + * Retrieves default power management limit on this device, in milliwatts. + * Default power management limit is a power management limit that the device boots with. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param defaultLimit Reference in which to return the default power management limit in milliwatts + * + * @return + * - \ref NVML_SUCCESS if \a defaultLimit has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a defaultLimit is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPowerManagementDefaultLimit(nvmlDevice_t device, unsigned int *defaultLimit); + +/** + * Retrieves power usage for this GPU in milliwatts and its associated circuitry (e.g. memory) + * + * For Fermi &tm; or newer fully supported devices. + * + * On Fermi and Kepler GPUs the reading is accurate to within +/- 5% of current power draw. On Ampere + * (except GA100) or newer GPUs, the API returns power averaged over 1 sec interval. On GA100 and + * older architectures, instantaneous power is returned. + * + * See \ref NVML_FI_DEV_POWER_AVERAGE and \ref NVML_FI_DEV_POWER_INSTANT to query specific power + * values. + * + * It is only available if power management mode is supported. See \ref nvmlDeviceGetPowerManagementMode. + * + * @param device The identifier of the target device + * @param power Reference in which to return the power usage information + * + * @return + * - \ref NVML_SUCCESS if \a power has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a power is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support power readings + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPowerUsage(nvmlDevice_t device, unsigned int *power); + +/** + * Retrieves total energy consumption for this GPU in millijoules (mJ) since the driver was last reloaded + * + * For Volta &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param energy Reference in which to return the energy consumption information + * + * @return + * - \ref NVML_SUCCESS if \a energy has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a energy is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support energy readings + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetTotalEnergyConsumption(nvmlDevice_t device, unsigned long long *energy); + +/** + * Get the effective power limit that the driver enforces after taking into account all limiters + * + * Note: This can be different from the \ref nvmlDeviceGetPowerManagementLimit if other limits are set elsewhere + * This includes the out of band power limit interface + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The device to communicate with + * @param limit Reference in which to return the power management limit in milliwatts + * + * @return + * - \ref NVML_SUCCESS if \a limit has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a limit is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetEnforcedPowerLimit(nvmlDevice_t device, unsigned int *limit); + +/** + * Retrieves the current GOM and pending GOM (the one that GPU will switch to after reboot). + * + * For GK110 M-class and X-class Tesla &tm; products from the Kepler family. + * Modes \ref NVML_GOM_LOW_DP and \ref NVML_GOM_ALL_ON are supported on fully supported GeForce products. + * Not supported on Quadro ® and Tesla &tm; C-class products. + * + * @param device The identifier of the target device + * @param current Reference in which to return the current GOM + * @param pending Reference in which to return the pending GOM + * + * @return + * - \ref NVML_SUCCESS if \a mode has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a current or \a pending is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlGpuOperationMode_t + * @see nvmlDeviceSetGpuOperationMode + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuOperationMode(nvmlDevice_t device, nvmlGpuOperationMode_t *current, nvmlGpuOperationMode_t *pending); + +/** + * Retrieves the amount of used, free, reserved and total memory available on the device, in bytes. + * The reserved amount is supported on version 2 only. + * + * For all products. + * + * Enabling ECC reduces the amount of total available memory, due to the extra required parity bits. + * Under WDDM most device memory is allocated and managed on startup by Windows. + * + * Under Linux and Windows TCC, the reported amount of used memory is equal to the sum of memory allocated + * by all active channels on the device. + * + * See \ref nvmlMemory_v2_t for details on available memory info. + * + * @note In MIG mode, if device handle is provided, the API returns aggregate + * information, only if the caller has appropriate privileges. Per-instance + * information can be queried by using specific MIG device handles. + * + * @note nvmlDeviceGetMemoryInfo_v2 adds additional memory information. + * + * @note On systems where GPUs are NUMA nodes, the accuracy of FB memory utilization + * provided by this API depends on the memory accounting of the operating system. + * This is because FB memory is managed by the operating system instead of the NVIDIA GPU driver. + * Typically, pages allocated from FB memory are not released even after + * the process terminates to enhance performance. In scenarios where + * the operating system is under memory pressure, it may resort to utilizing FB memory. + * Such actions can result in discrepancies in the accuracy of memory reporting. + * + * @param device The identifier of the target device + * @param memory Reference in which to return the memory information + * + * @return + * - \ref NVML_SUCCESS if \a memory has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a memory is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMemoryInfo(nvmlDevice_t device, nvmlMemory_t *memory); +nvmlReturn_t DECLDIR nvmlDeviceGetMemoryInfo_v2(nvmlDevice_t device, nvmlMemory_v2_t *memory); + +/** + * Retrieves the current compute mode for the device. + * + * For all products. + * + * See \ref nvmlComputeMode_t for details on allowed compute modes. + * + * @param device The identifier of the target device + * @param mode Reference in which to return the current compute mode + * + * @return + * - \ref NVML_SUCCESS if \a mode has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceSetComputeMode() + */ +nvmlReturn_t DECLDIR nvmlDeviceGetComputeMode(nvmlDevice_t device, nvmlComputeMode_t *mode); + +/** + * Retrieves the CUDA compute capability of the device. + * + * For all products. + * + * Returns the major and minor compute capability version numbers of the + * device. The major and minor versions are equivalent to the + * CU_DEVICE_ATTRIBUTE_COMPUTE_CAPABILITY_MINOR and + * CU_DEVICE_ATTRIBUTE_COMPUTE_CAPABILITY_MAJOR attributes that would be + * returned by CUDA's cuDeviceGetAttribute(). + * + * @param device The identifier of the target device + * @param major Reference in which to return the major CUDA compute capability + * @param minor Reference in which to return the minor CUDA compute capability + * + * @return + * - \ref NVML_SUCCESS if \a major and \a minor have been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a major or \a minor are NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetCudaComputeCapability(nvmlDevice_t device, int *major, int *minor); + +/** + * Retrieves the current and pending ECC modes for the device. + * + * For Fermi &tm; or newer fully supported devices. + * Only applicable to devices with ECC. + * Requires \a NVML_INFOROM_ECC version 1.0 or higher. + * + * Changing ECC modes requires a reboot. The "pending" ECC mode refers to the target mode following + * the next reboot. + * + * See \ref nvmlEnableState_t for details on allowed modes. + * + * @param device The identifier of the target device + * @param current Reference in which to return the current ECC mode + * @param pending Reference in which to return the pending ECC mode + * + * @return + * - \ref NVML_SUCCESS if \a current and \a pending have been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or either \a current or \a pending is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceSetEccMode() + */ +nvmlReturn_t DECLDIR nvmlDeviceGetEccMode(nvmlDevice_t device, nvmlEnableState_t *current, nvmlEnableState_t *pending); + +/** + * Retrieves the default ECC modes for the device. + * + * For Fermi &tm; or newer fully supported devices. + * Only applicable to devices with ECC. + * Requires \a NVML_INFOROM_ECC version 1.0 or higher. + * + * See \ref nvmlEnableState_t for details on allowed modes. + * + * @param device The identifier of the target device + * @param defaultMode Reference in which to return the default ECC mode + * + * @return + * - \ref NVML_SUCCESS if \a current and \a pending have been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a default is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceSetEccMode() + */ +nvmlReturn_t DECLDIR nvmlDeviceGetDefaultEccMode(nvmlDevice_t device, nvmlEnableState_t *defaultMode); + +/** + * Retrieves the device boardId from 0-N. + * Devices with the same boardId indicate GPUs connected to the same PLX. Use in conjunction with + * \ref nvmlDeviceGetMultiGpuBoard() to decide if they are on the same board as well. + * The boardId returned is a unique ID for the current configuration. Uniqueness and ordering across + * reboots and system configurations is not guaranteed (i.e. if a Tesla K40c returns 0x100 and + * the two GPUs on a Tesla K10 in the same system returns 0x200 it is not guaranteed they will + * always return those values but they will always be different from each other). + * + * + * For Fermi &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param boardId Reference in which to return the device's board ID + * + * @return + * - \ref NVML_SUCCESS if \a boardId has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a boardId is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetBoardId(nvmlDevice_t device, unsigned int *boardId); + +/** + * Retrieves whether the device is on a Multi-GPU Board + * Devices that are on multi-GPU boards will set \a multiGpuBool to a non-zero value. + * + * For Fermi &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param multiGpuBool Reference in which to return a zero or non-zero value + * to indicate whether the device is on a multi GPU board + * + * @return + * - \ref NVML_SUCCESS if \a multiGpuBool has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a multiGpuBool is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMultiGpuBoard(nvmlDevice_t device, unsigned int *multiGpuBool); + +/** + * Retrieves the total ECC error counts for the device. + * + * For Fermi &tm; or newer fully supported devices. + * Only applicable to devices with ECC. + * Requires \a NVML_INFOROM_ECC version 1.0 or higher. + * Requires ECC Mode to be enabled. + * + * The total error count is the sum of errors across each of the separate memory systems, i.e. the total set of + * errors across the entire device. + * + * See \ref nvmlMemoryErrorType_t for a description of available error types.\n + * See \ref nvmlEccCounterType_t for a description of available counter types. + * + * @param device The identifier of the target device + * @param errorType Flag that specifies the type of the errors. + * @param counterType Flag that specifies the counter-type of the errors. + * @param eccCounts Reference in which to return the specified ECC errors + * + * @return + * - \ref NVML_SUCCESS if \a eccCounts has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a errorType or \a counterType is invalid, or \a eccCounts is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceClearEccErrorCounts() + */ +nvmlReturn_t DECLDIR nvmlDeviceGetTotalEccErrors(nvmlDevice_t device, nvmlMemoryErrorType_t errorType, nvmlEccCounterType_t counterType, unsigned long long *eccCounts); + +/** + * Retrieves the detailed ECC error counts for the device. + * + * @deprecated This API supports only a fixed set of ECC error locations + * On different GPU architectures different locations are supported + * See \ref nvmlDeviceGetMemoryErrorCounter + * + * For Fermi &tm; or newer fully supported devices. + * Only applicable to devices with ECC. + * Requires \a NVML_INFOROM_ECC version 2.0 or higher to report aggregate location-based ECC counts. + * Requires \a NVML_INFOROM_ECC version 1.0 or higher to report all other ECC counts. + * Requires ECC Mode to be enabled. + * + * Detailed errors provide separate ECC counts for specific parts of the memory system. + * + * Reports zero for unsupported ECC error counters when a subset of ECC error counters are supported. + * + * See \ref nvmlMemoryErrorType_t for a description of available bit types.\n + * See \ref nvmlEccCounterType_t for a description of available counter types.\n + * See \ref nvmlEccErrorCounts_t for a description of provided detailed ECC counts. + * + * @param device The identifier of the target device + * @param errorType Flag that specifies the type of the errors. + * @param counterType Flag that specifies the counter-type of the errors. + * @param eccCounts Reference in which to return the specified ECC errors + * + * @return + * - \ref NVML_SUCCESS if \a eccCounts has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a errorType or \a counterType is invalid, or \a eccCounts is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceClearEccErrorCounts() + */ +nvmlReturn_t DECLDIR nvmlDeviceGetDetailedEccErrors(nvmlDevice_t device, nvmlMemoryErrorType_t errorType, nvmlEccCounterType_t counterType, nvmlEccErrorCounts_t *eccCounts); + +/** + * Retrieves the requested memory error counter for the device. + * + * For Fermi &tm; or newer fully supported devices. + * Requires \a NVML_INFOROM_ECC version 2.0 or higher to report aggregate location-based memory error counts. + * Requires \a NVML_INFOROM_ECC version 1.0 or higher to report all other memory error counts. + * + * Only applicable to devices with ECC. + * + * Requires ECC Mode to be enabled. + * + * @note On MIG-enabled GPUs, per instance information can be queried using specific + * MIG device handles. Per instance information is currently only supported for + * non-DRAM uncorrectable volatile errors. Querying volatile errors using device + * handles is currently not supported. + * + * See \ref nvmlMemoryErrorType_t for a description of available memory error types.\n + * See \ref nvmlEccCounterType_t for a description of available counter types.\n + * See \ref nvmlMemoryLocation_t for a description of available counter locations.\n + * + * @param device The identifier of the target device + * @param errorType Flag that specifies the type of error. + * @param counterType Flag that specifies the counter-type of the errors. + * @param locationType Specifies the location of the counter. + * @param count Reference in which to return the ECC counter + * + * @return + * - \ref NVML_SUCCESS if \a count has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a bitTyp,e \a counterType or \a locationType is + * invalid, or \a count is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support ECC error reporting in the specified memory + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMemoryErrorCounter(nvmlDevice_t device, nvmlMemoryErrorType_t errorType, + nvmlEccCounterType_t counterType, + nvmlMemoryLocation_t locationType, unsigned long long *count); + +/** + * Retrieves the current utilization rates for the device's major subsystems. + * + * For Fermi &tm; or newer fully supported devices. + * + * See \ref nvmlUtilization_t for details on available utilization rates. + * + * \note During driver initialization when ECC is enabled one can see high GPU and Memory Utilization readings. + * This is caused by ECC Memory Scrubbing mechanism that is performed during driver initialization. + * + * @note On MIG-enabled GPUs, querying device utilization rates is not currently supported. + * + * @param device The identifier of the target device + * @param utilization Reference in which to return the utilization information + * + * @return + * - \ref NVML_SUCCESS if \a utilization has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a utilization is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetUtilizationRates(nvmlDevice_t device, nvmlUtilization_t *utilization); + +/** + * Retrieves the current utilization and sampling size in microseconds for the Encoder + * + * For Kepler &tm; or newer fully supported devices. + * + * @note On MIG-enabled GPUs, querying encoder utilization is not currently supported. + * + * @param device The identifier of the target device + * @param utilization Reference to an unsigned int for encoder utilization info + * @param samplingPeriodUs Reference to an unsigned int for the sampling period in US + * + * @return + * - \ref NVML_SUCCESS if \a utilization has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a utilization is NULL, or \a samplingPeriodUs is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetEncoderUtilization(nvmlDevice_t device, unsigned int *utilization, unsigned int *samplingPeriodUs); + +/** + * Retrieves the current capacity of the device's encoder, as a percentage of maximum encoder capacity with valid values in the range 0-100. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param encoderQueryType Type of encoder to query + * @param encoderCapacity Reference to an unsigned int for the encoder capacity + * + * @return + * - \ref NVML_SUCCESS if \a encoderCapacity is fetched + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a encoderCapacity is NULL, or \a device or \a encoderQueryType + * are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if device does not support the encoder specified in \a encodeQueryType + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetEncoderCapacity (nvmlDevice_t device, nvmlEncoderType_t encoderQueryType, unsigned int *encoderCapacity); + +/** + * Retrieves the current encoder statistics for a given device. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param sessionCount Reference to an unsigned int for count of active encoder sessions + * @param averageFps Reference to an unsigned int for trailing average FPS of all active sessions + * @param averageLatency Reference to an unsigned int for encode latency in microseconds + * + * @return + * - \ref NVML_SUCCESS if \a sessionCount, \a averageFps and \a averageLatency is fetched + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a sessionCount, or \a device or \a averageFps, + * or \a averageLatency is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetEncoderStats (nvmlDevice_t device, unsigned int *sessionCount, + unsigned int *averageFps, unsigned int *averageLatency); + +/** + * Retrieves information about active encoder sessions on a target device. + * + * An array of active encoder sessions is returned in the caller-supplied buffer pointed at by \a sessionInfos. The + * array element count is passed in \a sessionCount, and \a sessionCount is used to return the number of sessions + * written to the buffer. + * + * If the supplied buffer is not large enough to accommodate the active session array, the function returns + * NVML_ERROR_INSUFFICIENT_SIZE, with the element count of nvmlEncoderSessionInfo_t array required in \a sessionCount. + * To query the number of active encoder sessions, call this function with *sessionCount = 0. The code will return + * NVML_SUCCESS with number of active encoder sessions updated in *sessionCount. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param sessionCount Reference to caller supplied array size, and returns the number of sessions. + * @param sessionInfos Reference in which to return the session information + * + * @return + * - \ref NVML_SUCCESS if \a sessionInfos is fetched + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a sessionCount is too small, array element count is returned in \a sessionCount + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a sessionCount is NULL. + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by \a device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetEncoderSessions(nvmlDevice_t device, unsigned int *sessionCount, nvmlEncoderSessionInfo_t *sessionInfos); + +/** + * Retrieves the current utilization and sampling size in microseconds for the Decoder + * + * For Kepler &tm; or newer fully supported devices. + * + * @note On MIG-enabled GPUs, querying decoder utilization is not currently supported. + * + * @param device The identifier of the target device + * @param utilization Reference to an unsigned int for decoder utilization info + * @param samplingPeriodUs Reference to an unsigned int for the sampling period in US + * + * @return + * - \ref NVML_SUCCESS if \a utilization has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a utilization is NULL, or \a samplingPeriodUs is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetDecoderUtilization(nvmlDevice_t device, unsigned int *utilization, unsigned int *samplingPeriodUs); + +/** + * Retrieves the current utilization and sampling size in microseconds for the JPG + * + * %TURING_OR_NEWER% + * + * @note On MIG-enabled GPUs, querying decoder utilization is not currently supported. + * + * @param device The identifier of the target device + * @param utilization Reference to an unsigned int for jpg utilization info + * @param samplingPeriodUs Reference to an unsigned int for the sampling period in US + * + * @return + * - \ref NVML_SUCCESS if \a utilization has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a utilization is NULL, or \a samplingPeriodUs is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetJpgUtilization(nvmlDevice_t device, unsigned int *utilization, unsigned int *samplingPeriodUs); + +/** + * Retrieves the current utilization and sampling size in microseconds for the OFA (Optical Flow Accelerator) + * + * %TURING_OR_NEWER% + * + * @note On MIG-enabled GPUs, querying decoder utilization is not currently supported. + * + * @param device The identifier of the target device + * @param utilization Reference to an unsigned int for ofa utilization info + * @param samplingPeriodUs Reference to an unsigned int for the sampling period in US + * + * @return + * - \ref NVML_SUCCESS if \a utilization has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a utilization is NULL, or \a samplingPeriodUs is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetOfaUtilization(nvmlDevice_t device, unsigned int *utilization, unsigned int *samplingPeriodUs); + +/** +* Retrieves the active frame buffer capture sessions statistics for a given device. +* +* For Maxwell &tm; or newer fully supported devices. +* +* @param device The identifier of the target device +* @param fbcStats Reference to nvmlFBCStats_t structure containing NvFBC stats +* +* @return +* - \ref NVML_SUCCESS if \a fbcStats is fetched +* - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized +* - \ref NVML_ERROR_INVALID_ARGUMENT if \a fbcStats is NULL +* - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible +* - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlDeviceGetFBCStats(nvmlDevice_t device, nvmlFBCStats_t *fbcStats); + +/** +* Retrieves information about active frame buffer capture sessions on a target device. +* +* An array of active FBC sessions is returned in the caller-supplied buffer pointed at by \a sessionInfo. The +* array element count is passed in \a sessionCount, and \a sessionCount is used to return the number of sessions +* written to the buffer. +* +* If the supplied buffer is not large enough to accommodate the active session array, the function returns +* NVML_ERROR_INSUFFICIENT_SIZE, with the element count of nvmlFBCSessionInfo_t array required in \a sessionCount. +* To query the number of active FBC sessions, call this function with *sessionCount = 0. The code will return +* NVML_SUCCESS with number of active FBC sessions updated in *sessionCount. +* +* For Maxwell &tm; or newer fully supported devices. +* +* @note hResolution, vResolution, averageFPS and averageLatency data for a FBC session returned in \a sessionInfo may +* be zero if there are no new frames captured since the session started. +* +* @param device The identifier of the target device +* @param sessionCount Reference to caller supplied array size, and returns the number of sessions. +* @param sessionInfo Reference in which to return the session information +* +* @return +* - \ref NVML_SUCCESS if \a sessionInfo is fetched +* - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized +* - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a sessionCount is too small, array element count is returned in \a sessionCount +* - \ref NVML_ERROR_INVALID_ARGUMENT if \a sessionCount is NULL. +* - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible +* - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlDeviceGetFBCSessions(nvmlDevice_t device, unsigned int *sessionCount, nvmlFBCSessionInfo_t *sessionInfo); + +/** + * Retrieves the current and pending driver model for the device. + * + * For Fermi &tm; or newer fully supported devices. + * For windows only. + * + * On Windows platforms the device driver can run in either WDDM or WDM (TCC) mode. If a display is attached + * to the device it must run in WDDM mode. TCC mode is preferred if a display is not attached. + * + * See \ref nvmlDriverModel_t for details on available driver models. + * + * @param device The identifier of the target device + * @param current Reference in which to return the current driver model + * @param pending Reference in which to return the pending driver model + * + * @return + * - \ref NVML_SUCCESS if either \a current and/or \a pending have been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or both \a current and \a pending are NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the platform is not windows + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceSetDriverModel() + */ +nvmlReturn_t DECLDIR nvmlDeviceGetDriverModel(nvmlDevice_t device, nvmlDriverModel_t *current, nvmlDriverModel_t *pending); + +/** + * Get VBIOS version of the device. + * + * For all products. + * + * The VBIOS version may change from time to time. It will not exceed 32 characters in length + * (including the NULL terminator). See \ref nvmlConstants::NVML_DEVICE_VBIOS_VERSION_BUFFER_SIZE. + * + * @param device The identifier of the target device + * @param version Reference to which to return the VBIOS version + * @param length The maximum allowed length of the string returned in \a version + * + * @return + * - \ref NVML_SUCCESS if \a version has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a version is NULL + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVbiosVersion(nvmlDevice_t device, char *version, unsigned int length); + +/** + * Get Bridge Chip Information for all the bridge chips on the board. + * + * For all fully supported products. + * Only applicable to multi-GPU products. + * + * @param device The identifier of the target device + * @param bridgeHierarchy Reference to the returned bridge chip Hierarchy + * + * @return + * - \ref NVML_SUCCESS if bridge chip exists + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a bridgeInfo is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if bridge chip not supported on the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetBridgeChipInfo(nvmlDevice_t device, nvmlBridgeChipHierarchy_t *bridgeHierarchy); + +/** + * Get information about processes with a compute context on a device + * + * For Fermi &tm; or newer fully supported devices. + * + * This function returns information only about compute running processes (e.g. CUDA application which have + * active context). Any graphics applications (e.g. using OpenGL, DirectX) won't be listed by this function. + * + * To query the current number of running compute processes, call this function with *infoCount = 0. The + * return code will be NVML_ERROR_INSUFFICIENT_SIZE, or NVML_SUCCESS if none are running. For this call + * \a infos is allowed to be NULL. + * + * The usedGpuMemory field returned is all of the memory used by the application. + * + * Keep in mind that information returned by this call is dynamic and the number of elements might change in + * time. Allocate more space for \a infos table in case new compute processes are spawned. + * + * @note In MIG mode, if device handle is provided, the API returns aggregate information, only if + * the caller has appropriate privileges. Per-instance information can be queried by using + * specific MIG device handles. + * Querying per-instance information using MIG device handles is not supported if the device is in vGPU Host virtualization mode. + * + * @param device The device handle or MIG device handle + * @param infoCount Reference in which to provide the \a infos array size, and + * to return the number of returned elements + * @param infos Reference in which to return the process information + * + * @return + * - \ref NVML_SUCCESS if \a infoCount and \a infos have been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a infoCount indicates that the \a infos array is too small + * \a infoCount will contain minimal amount of space necessary for + * the call to complete + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, either of \a infoCount or \a infos is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by \a device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see \ref nvmlSystemGetProcessName + */ +nvmlReturn_t DECLDIR nvmlDeviceGetComputeRunningProcesses_v3(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_t *infos); + +/** + * Get information about processes with a graphics context on a device + * + * For Kepler &tm; or newer fully supported devices. + * + * This function returns information only about graphics based processes + * (eg. applications using OpenGL, DirectX) + * + * To query the current number of running graphics processes, call this function with *infoCount = 0. The + * return code will be NVML_ERROR_INSUFFICIENT_SIZE, or NVML_SUCCESS if none are running. For this call + * \a infos is allowed to be NULL. + * + * The usedGpuMemory field returned is all of the memory used by the application. + * + * Keep in mind that information returned by this call is dynamic and the number of elements might change in + * time. Allocate more space for \a infos table in case new graphics processes are spawned. + * + * @note In MIG mode, if device handle is provided, the API returns aggregate information, only if + * the caller has appropriate privileges. Per-instance information can be queried by using + * specific MIG device handles. + * Querying per-instance information using MIG device handles is not supported if the device is in vGPU Host virtualization mode. + * + * @param device The device handle or MIG device handle + * @param infoCount Reference in which to provide the \a infos array size, and + * to return the number of returned elements + * @param infos Reference in which to return the process information + * + * @return + * - \ref NVML_SUCCESS if \a infoCount and \a infos have been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a infoCount indicates that the \a infos array is too small + * \a infoCount will contain minimal amount of space necessary for + * the call to complete + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, either of \a infoCount or \a infos is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by \a device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see \ref nvmlSystemGetProcessName + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGraphicsRunningProcesses_v3(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_t *infos); + +/** + * Get information about processes with a MPS compute context on a device + * + * For Volta &tm; or newer fully supported devices. + * + * This function returns information only about compute running processes (e.g. CUDA application which have + * active context) utilizing MPS. Any graphics applications (e.g. using OpenGL, DirectX) won't be listed by + * this function. + * + * To query the current number of running compute processes, call this function with *infoCount = 0. The + * return code will be NVML_ERROR_INSUFFICIENT_SIZE, or NVML_SUCCESS if none are running. For this call + * \a infos is allowed to be NULL. + * + * The usedGpuMemory field returned is all of the memory used by the application. + * + * Keep in mind that information returned by this call is dynamic and the number of elements might change in + * time. Allocate more space for \a infos table in case new compute processes are spawned. + * + * @note In MIG mode, if device handle is provided, the API returns aggregate information, only if + * the caller has appropriate privileges. Per-instance information can be queried by using + * specific MIG device handles. + * Querying per-instance information using MIG device handles is not supported if the device is in vGPU Host virtualization mode. + * + * @param device The device handle or MIG device handle + * @param infoCount Reference in which to provide the \a infos array size, and + * to return the number of returned elements + * @param infos Reference in which to return the process information + * + * @return + * - \ref NVML_SUCCESS if \a infoCount and \a infos have been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a infoCount indicates that the \a infos array is too small + * \a infoCount will contain minimal amount of space necessary for + * the call to complete + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, either of \a infoCount or \a infos is NULL + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by \a device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see \ref nvmlSystemGetProcessName + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMPSComputeRunningProcesses_v3(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_t *infos); + +/** + * Get information about running processes on a device for input context + * + * %HOPPER_OR_NEWER% + * + * This function returns information only about running processes (e.g. CUDA application which have + * active context). + * + * To determine the size of the @ref plist->procArray array to allocate, call the function with + * @ref plist->numProcArrayEntries set to zero and @ref plist->procArray set to NULL. The return + * code will be either NVML_ERROR_INSUFFICIENT_SIZE (if there are valid processes of type + * @ref plist->mode to report on, in which case the @ref plist->numProcArrayEntries field will + * indicate the required number of entries in the array) or NVML_SUCCESS (if no processes of type + * @ref plist->mode exist). + * + * The usedGpuMemory field returned is all of the memory used by the application. + * The usedGpuCcProtectedMemory field returned is all of the protected memory used by the application. + * + * Keep in mind that information returned by this call is dynamic and the number of elements might change in + * time. Allocate more space for \a plist->procArray table in case new processes are spawned. + * + * @note In MIG mode, if device handle is provided, the API returns aggregate information, only if + * the caller has appropriate privileges. Per-instance information can be queried by using + * specific MIG device handles. + * Querying per-instance information using MIG device handles is not supported if the device is in + * vGPU Host virtualization mode. + * Protected memory usage is currently not available in MIG mode and in windows. + * + * @param device The device handle or MIG device handle + * @param plist Reference in which to process detail list + * @param plist->version The api version + * @param plist->mode The process mode + * @param plist->procArray Reference in which to return the process information + * @param plist->numProcArrayEntries Proc array size of returned entries + * + * @return + * - \ref NVML_SUCCESS if \a plist->numprocArrayEntries and \a plist->procArray have been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a plist->numprocArrayEntries indicates that the \a plist->procArray is too small + * \a plist->numprocArrayEntries will contain minimal amount of space necessary for + * the call to complete + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a plist is NULL, \a plist->version is invalid, + * \a plist->mode is invalid, + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by \a device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetRunningProcessDetailList(nvmlDevice_t device, nvmlProcessDetailList_t *plist); + +/** + * Check if the GPU devices are on the same physical board. + * + * For all fully supported products. + * + * @param device1 The first GPU device + * @param device2 The second GPU device + * @param onSameBoard Reference in which to return the status. + * Non-zero indicates that the GPUs are on the same board. + * + * @return + * - \ref NVML_SUCCESS if \a onSameBoard has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a dev1 or \a dev2 are invalid or \a onSameBoard is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this check is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the either GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceOnSameBoard(nvmlDevice_t device1, nvmlDevice_t device2, int *onSameBoard); + +/** + * Retrieves the root/admin permissions on the target API. See \a nvmlRestrictedAPI_t for the list of supported APIs. + * If an API is restricted only root users can call that API. See \a nvmlDeviceSetAPIRestriction to change current permissions. + * + * For all fully supported products. + * + * @param device The identifier of the target device + * @param apiType Target API type for this operation + * @param isRestricted Reference in which to return the current restriction + * NVML_FEATURE_ENABLED indicates that the API is root-only + * NVML_FEATURE_DISABLED indicates that the API is accessible to all users + * + * @return + * - \ref NVML_SUCCESS if \a isRestricted has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a apiType incorrect or \a isRestricted is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device or the device does not support + * the feature that is being queried (E.G. Enabling/disabling Auto Boosted clocks is + * not supported by the device) + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlRestrictedAPI_t + */ +nvmlReturn_t DECLDIR nvmlDeviceGetAPIRestriction(nvmlDevice_t device, nvmlRestrictedAPI_t apiType, nvmlEnableState_t *isRestricted); + +/** + * Gets recent samples for the GPU. + * + * For Kepler &tm; or newer fully supported devices. + * + * Based on type, this method can be used to fetch the power, utilization or clock samples maintained in the buffer by + * the driver. + * + * Power, Utilization and Clock samples are returned as type "unsigned int" for the union nvmlValue_t. + * + * To get the size of samples that user needs to allocate, the method is invoked with samples set to NULL. + * The returned samplesCount will provide the number of samples that can be queried. The user needs to + * allocate the buffer with size as samplesCount * sizeof(nvmlSample_t). + * + * lastSeenTimeStamp represents CPU timestamp in microseconds. Set it to 0 to fetch all the samples maintained by the + * underlying buffer. Set lastSeenTimeStamp to one of the timeStamps retrieved from the date of the previous query + * to get more recent samples. + * + * This method fetches the number of entries which can be accommodated in the provided samples array, and the + * reference samplesCount is updated to indicate how many samples were actually retrieved. The advantage of using this + * method for samples in contrast to polling via existing methods is to get get higher frequency data at lower polling cost. + * + * @note On MIG-enabled GPUs, querying the following sample types, NVML_GPU_UTILIZATION_SAMPLES, NVML_MEMORY_UTILIZATION_SAMPLES + * NVML_ENC_UTILIZATION_SAMPLES and NVML_DEC_UTILIZATION_SAMPLES, is not currently supported. + * + * @param device The identifier for the target device + * @param type Type of sampling event + * @param lastSeenTimeStamp Return only samples with timestamp greater than lastSeenTimeStamp. + * @param sampleValType Output parameter to represent the type of sample value as described in nvmlSampleVal_t + * @param sampleCount Reference to provide the number of elements which can be queried in samples array + * @param samples Reference in which samples are returned + + * @return + * - \ref NVML_SUCCESS if samples are successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a samplesCount is NULL or + * reference to \a sampleCount is 0 for non null \a samples + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_NOT_FOUND if sample entries are not found + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetSamples(nvmlDevice_t device, nvmlSamplingType_t type, unsigned long long lastSeenTimeStamp, + nvmlValueType_t *sampleValType, unsigned int *sampleCount, nvmlSample_t *samples); + +/** + * Gets Total, Available and Used size of BAR1 memory. + * + * BAR1 is used to map the FB (device memory) so that it can be directly accessed by the CPU or by 3rd party + * devices (peer-to-peer on the PCIE bus). + * + * @note In MIG mode, if device handle is provided, the API returns aggregate + * information, only if the caller has appropriate privileges. Per-instance + * information can be queried by using specific MIG device handles. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param bar1Memory Reference in which BAR1 memory + * information is returned. + * + * @return + * - \ref NVML_SUCCESS if BAR1 memory is successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a bar1Memory is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetBAR1MemoryInfo(nvmlDevice_t device, nvmlBAR1Memory_t *bar1Memory); + +/** + * Gets the duration of time during which the device was throttled (lower than requested clocks) due to power + * or thermal constraints. + * + * The method is important to users who are tying to understand if their GPUs throttle at any point during their applications. The + * difference in violation times at two different reference times gives the indication of GPU throttling event. + * + * Violation for thermal capping is not supported at this time. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param perfPolicyType Represents Performance policy which can trigger GPU throttling + * @param violTime Reference to which violation time related information is returned + * + * + * @return + * - \ref NVML_SUCCESS if violation time is successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a perfPolicyType is invalid, or \a violTime is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetViolationStatus(nvmlDevice_t device, nvmlPerfPolicyType_t perfPolicyType, nvmlViolationTime_t *violTime); + +/** + * Gets the device's interrupt number + * + * @param device The identifier of the target device + * @param irqNum The interrupt number associated with the specified device + * + * @return + * - \ref NVML_SUCCESS if irq number is successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a irqNum is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetIrqNum(nvmlDevice_t device, unsigned int *irqNum); + +/** + * Gets the device's core count + * + * @param device The identifier of the target device + * @param numCores The number of cores for the specified device + * + * @return + * - \ref NVML_SUCCESS if Gpu core count is successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a numCores is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetNumGpuCores(nvmlDevice_t device, unsigned int *numCores); + +/** + * Gets the devices power source + * + * @param device The identifier of the target device + * @param powerSource The power source of the device + * + * @return + * - \ref NVML_SUCCESS if the current power source was successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a powerSource is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPowerSource(nvmlDevice_t device, nvmlPowerSource_t *powerSource); + +/** + * Gets the device's memory bus width + * + * @param device The identifier of the target device + * @param busWidth The devices's memory bus width + * + * @return + * - \ref NVML_SUCCESS if the memory bus width is successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a busWidth is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMemoryBusWidth(nvmlDevice_t device, unsigned int *busWidth); + +/** + * Gets the device's PCIE Max Link speed in MBPS + * + * @param device The identifier of the target device + * @param maxSpeed The devices's PCIE Max Link speed in MBPS + * + * @return + * - \ref NVML_SUCCESS if Pcie Max Link Speed is successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a maxSpeed is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPcieLinkMaxSpeed(nvmlDevice_t device, unsigned int *maxSpeed); + +/** + * Gets the device's PCIe Link speed in Mbps + * + * @param device The identifier of the target device + * @param pcieSpeed The devices's PCIe Max Link speed in Mbps + * + * @return + * - \ref NVML_SUCCESS if \a pcieSpeed has been retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pcieSpeed is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support PCIe speed getting + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPcieSpeed(nvmlDevice_t device, unsigned int *pcieSpeed); + +/** + * Gets the device's Adaptive Clock status + * + * @param device The identifier of the target device + * @param adaptiveClockStatus The current adaptive clocking status, either + * @ref NVML_ADAPTIVE_CLOCKING_INFO_STATUS_DISABLED + * or @ref NVML_ADAPTIVE_CLOCKING_INFO_STATUS_ENABLED + * + * @return + * - \ref NVML_SUCCESS if the current adaptive clocking status is successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a adaptiveClockStatus is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * + */ +nvmlReturn_t DECLDIR nvmlDeviceGetAdaptiveClockInfoStatus(nvmlDevice_t device, unsigned int *adaptiveClockStatus); + +/** + * Get the type of the GPU Bus (PCIe, PCI, ...) + * + * @param device The identifier of the target device + * @param type The PCI Bus type + * + * return + * - \ref NVML_SUCCESS if the bus \a type is successfully retreived + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \device is invalid or \type is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetBusType(nvmlDevice_t device, nvmlBusType_t *type); + +/** + * Get fabric information associated with the device. + * + * %HOPPER_OR_NEWER% + * + * On Hopper + NVSwitch systems, GPU is registered with the NVIDIA Fabric Manager + * Upon successful registration, the GPU is added to the NVLink fabric to enable + * peer-to-peer communication. + * This API reports the current state of the GPU in the NVLink fabric + * along with other useful information. + * + * @param device The identifier of the target device + * @param gpuFabricInfo Information about GPU fabric state + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't support gpu fabric + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuFabricInfo(nvmlDevice_t device, nvmlGpuFabricInfo_t *gpuFabricInfo); + +/** + * Get Conf Computing System capabilities. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param capabilities System CC capabilities + * + * @return + * - \ref NVML_SUCCESS if \a capabilities were successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a capabilities is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + */ +nvmlReturn_t DECLDIR nvmlSystemGetConfComputeCapabilities(nvmlConfComputeSystemCaps_t *capabilities); + +/** + * Get Conf Computing System State. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param state System CC State + * + * @return + * - \ref NVML_SUCCESS if \a state were successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a state is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + */ +nvmlReturn_t DECLDIR nvmlSystemGetConfComputeState(nvmlConfComputeSystemState_t *state); + +/** + * Get Conf Computing Protected and Unprotected Memory Sizes. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param device Device handle + * @param memInfo Protected/Unprotected Memory sizes + * + * @return + * - \ref NVML_SUCCESS if \a memInfo were successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a memInfo or \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + */ +nvmlReturn_t DECLDIR nvmlDeviceGetConfComputeMemSizeInfo(nvmlDevice_t device, nvmlConfComputeMemSizeInfo_t *memInfo); + +/** + * Get Conf Computing GPUs ready state. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param isAcceptingWork Returns GPU current work accepting state, + * NVML_CC_ACCEPTING_CLIENT_REQUESTS_TRUE or + * NVML_CC_ACCEPTING_CLIENT_REQUESTS_FALSE + * + * return + * - \ref NVML_SUCCESS if \a current GPUs ready state were successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a isAcceptingWork is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + */ +nvmlReturn_t DECLDIR nvmlSystemGetConfComputeGpusReadyState(unsigned int *isAcceptingWork); + +/** + * Get Conf Computing protected memory usage. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param device The identifier of the target device + * @param memory Reference in which to return the memory information + * + * @return + * - \ref NVML_SUCCESS if \a memory has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a memory is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetConfComputeProtectedMemoryUsage(nvmlDevice_t device, nvmlMemory_t *memory); + +/** + * Get Conf Computing Gpu certificate details. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param device The identifier of the target device + * @param gpuCert Reference in which to return the gpu certificate information + * + * @return + * - \ref NVML_SUCCESS if \a gpu certificate info has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a memory is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetConfComputeGpuCertificate(nvmlDevice_t device, + nvmlConfComputeGpuCertificate_t *gpuCert); + +/** + * Get Conf Computing Gpu attestation report. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param device The identifier of the target device + * @param gpuAtstReport Reference in which to return the gpu attestation report + * + * @return + * - \ref NVML_SUCCESS if \a gpu attestation report has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a memory is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetConfComputeGpuAttestationReport(nvmlDevice_t device, + nvmlConfComputeGpuAttestationReport_t *gpuAtstReport); + +/** + * Retrieve GSP firmware version. + * + * The caller passes in buffer via \a version and corresponding GSP firmware numbered version + * is returned with the same parameter in string format. + * + * @param device Device handle + * @param version The retrieved GSP firmware version + * + * @return + * - \ref NVML_SUCCESS if GSP firmware version is sucessfully retrieved + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or GSP \a version pointer is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if GSP firmware is not enabled for GPU + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGspFirmwareVersion(nvmlDevice_t device, char *version); + +/** + * Retrieve GSP firmware mode. + * + * The caller passes in integer pointers. GSP firmware enablement and default mode information is returned with + * corresponding parameters. The return value in \a isEnabled and \a defaultMode should be treated as boolean. + * + * @param device Device handle + * @param isEnabled Pointer to specify if GSP firmware is enabled + * @param defaultMode Pointer to specify if GSP firmware is supported by default on \a device + * + * @return + * - \ref NVML_SUCCESS if GSP firmware mode is sucessfully retrieved + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or any of \a isEnabled or \a defaultMode is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGspFirmwareMode(nvmlDevice_t device, unsigned int *isEnabled, unsigned int *defaultMode); + +/** + * @} + */ + +/** @addtogroup nvmlAccountingStats + * @{ + */ + +/** + * Queries the state of per process accounting mode. + * + * For Kepler &tm; or newer fully supported devices. + * + * See \ref nvmlDeviceGetAccountingStats for more details. + * See \ref nvmlDeviceSetAccountingMode + * + * @param device The identifier of the target device + * @param mode Reference in which to return the current accounting mode + * + * @return + * - \ref NVML_SUCCESS if the mode has been successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode are NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetAccountingMode(nvmlDevice_t device, nvmlEnableState_t *mode); + +/** + * Queries process's accounting stats. + * + * For Kepler &tm; or newer fully supported devices. + * + * Accounting stats capture GPU utilization and other statistics across the lifetime of a process. + * Accounting stats can be queried during life time of the process and after its termination. + * The time field in \ref nvmlAccountingStats_t is reported as 0 during the lifetime of the process and + * updated to actual running time after its termination. + * Accounting stats are kept in a circular buffer, newly created processes overwrite information about old + * processes. + * + * See \ref nvmlAccountingStats_t for description of each returned metric. + * List of processes that can be queried can be retrieved from \ref nvmlDeviceGetAccountingPids. + * + * @note Accounting Mode needs to be on. See \ref nvmlDeviceGetAccountingMode. + * @note Only compute and graphics applications stats can be queried. Monitoring applications stats can't be + * queried since they don't contribute to GPU utilization. + * @note In case of pid collision stats of only the latest process (that terminated last) will be reported + * + * @warning On Kepler devices per process statistics are accurate only if there's one process running on a GPU. + * + * @param device The identifier of the target device + * @param pid Process Id of the target process to query stats for + * @param stats Reference in which to return the process's accounting stats + * + * @return + * - \ref NVML_SUCCESS if stats have been successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a stats are NULL + * - \ref NVML_ERROR_NOT_FOUND if process stats were not found + * - \ref NVML_ERROR_NOT_SUPPORTED if \a device doesn't support this feature or accounting mode is disabled + * or on vGPU host. + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetAccountingBufferSize + */ +nvmlReturn_t DECLDIR nvmlDeviceGetAccountingStats(nvmlDevice_t device, unsigned int pid, nvmlAccountingStats_t *stats); + +/** + * Queries list of processes that can be queried for accounting stats. The list of processes returned + * can be in running or terminated state. + * + * For Kepler &tm; or newer fully supported devices. + * + * To just query the number of processes ready to be queried, call this function with *count = 0 and + * pids=NULL. The return code will be NVML_ERROR_INSUFFICIENT_SIZE, or NVML_SUCCESS if list is empty. + * + * For more details see \ref nvmlDeviceGetAccountingStats. + * + * @note In case of PID collision some processes might not be accessible before the circular buffer is full. + * + * @param device The identifier of the target device + * @param count Reference in which to provide the \a pids array size, and + * to return the number of elements ready to be queried + * @param pids Reference in which to return list of process ids + * + * @return + * - \ref NVML_SUCCESS if pids were successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a count is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if \a device doesn't support this feature or accounting mode is disabled + * or on vGPU host. + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a count is too small (\a count is set to + * expected value) + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetAccountingBufferSize + */ +nvmlReturn_t DECLDIR nvmlDeviceGetAccountingPids(nvmlDevice_t device, unsigned int *count, unsigned int *pids); + +/** + * Returns the number of processes that the circular buffer with accounting pids can hold. + * + * For Kepler &tm; or newer fully supported devices. + * + * This is the maximum number of processes that accounting information will be stored for before information + * about oldest processes will get overwritten by information about new processes. + * + * @param device The identifier of the target device + * @param bufferSize Reference in which to provide the size (in number of elements) + * of the circular buffer for accounting stats. + * + * @return + * - \ref NVML_SUCCESS if buffer size was successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a bufferSize is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature or accounting mode is disabled + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetAccountingStats + * @see nvmlDeviceGetAccountingPids + */ +nvmlReturn_t DECLDIR nvmlDeviceGetAccountingBufferSize(nvmlDevice_t device, unsigned int *bufferSize); + +/** @} */ + +/** @addtogroup nvmlDeviceQueries + * @{ + */ + +/** + * Returns the list of retired pages by source, including pages that are pending retirement + * The address information provided from this API is the hardware address of the page that was retired. Note + * that this does not match the virtual address used in CUDA, but will match the address information in XID 63 + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param cause Filter page addresses by cause of retirement + * @param pageCount Reference in which to provide the \a addresses buffer size, and + * to return the number of retired pages that match \a cause + * Set to 0 to query the size without allocating an \a addresses buffer + * @param addresses Buffer to write the page addresses into + * + * @return + * - \ref NVML_SUCCESS if \a pageCount was populated and \a addresses was filled + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a pageCount indicates the buffer is not large enough to store all the + * matching page addresses. \a pageCount is set to the needed size. + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a pageCount is NULL, \a cause is invalid, or + * \a addresses is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetRetiredPages(nvmlDevice_t device, nvmlPageRetirementCause_t cause, + unsigned int *pageCount, unsigned long long *addresses); + +/** + * Returns the list of retired pages by source, including pages that are pending retirement + * The address information provided from this API is the hardware address of the page that was retired. Note + * that this does not match the virtual address used in CUDA, but will match the address information in XID 63 + * + * \note nvmlDeviceGetRetiredPages_v2 adds an additional timestamps parameter to return the time of each page's + * retirement. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param cause Filter page addresses by cause of retirement + * @param pageCount Reference in which to provide the \a addresses buffer size, and + * to return the number of retired pages that match \a cause + * Set to 0 to query the size without allocating an \a addresses buffer + * @param addresses Buffer to write the page addresses into + * @param timestamps Buffer to write the timestamps of page retirement, additional for _v2 + * + * @return + * - \ref NVML_SUCCESS if \a pageCount was populated and \a addresses was filled + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a pageCount indicates the buffer is not large enough to store all the + * matching page addresses. \a pageCount is set to the needed size. + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a pageCount is NULL, \a cause is invalid, or + * \a addresses is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetRetiredPages_v2(nvmlDevice_t device, nvmlPageRetirementCause_t cause, + unsigned int *pageCount, unsigned long long *addresses, unsigned long long *timestamps); + +/** + * Check if any pages are pending retirement and need a reboot to fully retire. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param isPending Reference in which to return the pending status + * + * @return + * - \ref NVML_SUCCESS if \a isPending was populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a isPending is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetRetiredPagesPendingStatus(nvmlDevice_t device, nvmlEnableState_t *isPending); + +/** + * Get number of remapped rows. The number of rows reported will be based on + * the cause of the remapping. isPending indicates whether or not there are + * pending remappings. A reset will be required to actually remap the row. + * failureOccurred will be set if a row remapping ever failed in the past. A + * pending remapping won't affect future work on the GPU since + * error-containment and dynamic page blacklisting will take care of that. + * + * @note On MIG-enabled GPUs with active instances, querying the number of + * remapped rows is not supported + * + * For Ampere &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param corrRows Reference for number of rows remapped due to correctable errors + * @param uncRows Reference for number of rows remapped due to uncorrectable errors + * @param isPending Reference for whether or not remappings are pending + * @param failureOccurred Reference that is set when a remapping has failed in the past + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a corrRows, \a uncRows, \a isPending or \a failureOccurred is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If MIG is enabled or if the device doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN Unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetRemappedRows(nvmlDevice_t device, unsigned int *corrRows, unsigned int *uncRows, + unsigned int *isPending, unsigned int *failureOccurred); + +/** + * Get the row remapper histogram. Returns the remap availability for each bank + * on the GPU. + * + * @param device Device handle + * @param values Histogram values + * + * @return + * - \ref NVML_SUCCESS On success + * - \ref NVML_ERROR_UNKNOWN On any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetRowRemapperHistogram(nvmlDevice_t device, nvmlRowRemapperHistogramValues_t *values); + +/** + * Get architecture for device + * + * @param device The identifier of the target device + * @param arch Reference where architecture is returned, if call successful. + * Set to NVML_DEVICE_ARCH_* upon success + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device or \a arch (output refererence) are invalid + */ +nvmlReturn_t DECLDIR nvmlDeviceGetArchitecture(nvmlDevice_t device, nvmlDeviceArchitecture_t *arch); + +/** + * Retrieves the frequency monitor fault status for the device. + * + * For Ampere &tm; or newer fully supported devices. + * Requires root user. + * + * See \ref nvmlClkMonStatus_t for details on decoding the status output. + * + * @param device The identifier of the target device + * @param status Reference in which to return the clkmon fault status + * + * @return + * - \ref NVML_SUCCESS if \a status has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a status is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetClkMonStatus() + */ +nvmlReturn_t DECLDIR nvmlDeviceGetClkMonStatus(nvmlDevice_t device, nvmlClkMonStatus_t *status); + +/** + * Retrieves the current utilization and process ID + * + * For Maxwell &tm; or newer fully supported devices. + * + * Reads recent utilization of GPU SM (3D/Compute), framebuffer, video encoder, and video decoder for processes running. + * Utilization values are returned as an array of utilization sample structures in the caller-supplied buffer pointed at + * by \a utilization. One utilization sample structure is returned per process running, that had some non-zero utilization + * during the last sample period. It includes the CPU timestamp at which the samples were recorded. Individual utilization values + * are returned as "unsigned int" values. If no valid sample entries are found since the lastSeenTimeStamp, NVML_ERROR_NOT_FOUND + * is returned. + * + * To read utilization values, first determine the size of buffer required to hold the samples by invoking the function with + * \a utilization set to NULL. The caller should allocate a buffer of size + * processSamplesCount * sizeof(nvmlProcessUtilizationSample_t). Invoke the function again with the allocated buffer passed + * in \a utilization, and \a processSamplesCount set to the number of entries the buffer is sized for. + * + * On successful return, the function updates \a processSamplesCount with the number of process utilization sample + * structures that were actually written. This may differ from a previously read value as instances are created or + * destroyed. + * + * lastSeenTimeStamp represents the CPU timestamp in microseconds at which utilization samples were last read. Set it to 0 + * to read utilization based on all the samples maintained by the driver's internal sample buffer. Set lastSeenTimeStamp + * to a timeStamp retrieved from a previous query to read utilization since the previous query. + * + * @note On MIG-enabled GPUs, querying process utilization is not currently supported. + * + * @param device The identifier of the target device + * @param utilization Pointer to caller-supplied buffer in which guest process utilization samples are returned + * @param processSamplesCount Pointer to caller-supplied array size, and returns number of processes running + * @param lastSeenTimeStamp Return only samples with timestamp greater than lastSeenTimeStamp. + + * @return + * - \ref NVML_SUCCESS if \a utilization has been populated + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a utilization is NULL, or \a samplingPeriodUs is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_NOT_FOUND if sample entries are not found + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetProcessUtilization(nvmlDevice_t device, nvmlProcessUtilizationSample_t *utilization, + unsigned int *processSamplesCount, unsigned long long lastSeenTimeStamp); + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlUnitCommands Unit Commands + * This chapter describes NVML operations that change the state of the unit. For S-class products. + * Each of these requires root/admin access. Non-admin users will see an NVML_ERROR_NO_PERMISSION + * error code when invoking any of these methods. + * @{ + */ +/***************************************************************************************************/ + +/** + * Set the LED state for the unit. The LED can be either green (0) or amber (1). + * + * For S-class products. + * Requires root/admin permissions. + * + * This operation takes effect immediately. + * + * + * Current S-Class products don't provide unique LEDs for each unit. As such, both front + * and back LEDs will be toggled in unison regardless of which unit is specified with this command. + * + * See \ref nvmlLedColor_t for available colors. + * + * @param unit The identifier of the target unit + * @param color The target LED color + * + * @return + * - \ref NVML_SUCCESS if the LED color has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a unit or \a color is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this is not an S-class product + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlUnitGetLedState() + */ +nvmlReturn_t DECLDIR nvmlUnitSetLedState(nvmlUnit_t unit, nvmlLedColor_t color); + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlDeviceCommands Device Commands + * This chapter describes NVML operations that change the state of the device. + * Each of these requires root/admin access. Non-admin users will see an NVML_ERROR_NO_PERMISSION + * error code when invoking any of these methods. + * @{ + */ +/***************************************************************************************************/ + +/** + * Set the persistence mode for the device. + * + * For all products. + * For Linux only. + * Requires root/admin permissions. + * + * The persistence mode determines whether the GPU driver software is torn down after the last client + * exits. + * + * This operation takes effect immediately. It is not persistent across reboots. After each reboot the + * persistence mode is reset to "Disabled". + * + * See \ref nvmlEnableState_t for available modes. + * + * After calling this API with mode set to NVML_FEATURE_DISABLED on a device that has its own NUMA + * memory, the given device handle will no longer be valid, and to continue to interact with this + * device, a new handle should be obtained from one of the nvmlDeviceGetHandleBy*() APIs. This + * limitation is currently only applicable to devices that have a coherent NVLink connection to + * system memory. + * + * @param device The identifier of the target device + * @param mode The target persistence mode + * + * @return + * - \ref NVML_SUCCESS if the persistence mode was set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetPersistenceMode() + */ +nvmlReturn_t DECLDIR nvmlDeviceSetPersistenceMode(nvmlDevice_t device, nvmlEnableState_t mode); + +/** + * Set the compute mode for the device. + * + * For all products. + * Requires root/admin permissions. + * + * The compute mode determines whether a GPU can be used for compute operations and whether it can + * be shared across contexts. + * + * This operation takes effect immediately. Under Linux it is not persistent across reboots and + * always resets to "Default". Under windows it is persistent. + * + * Under windows compute mode may only be set to DEFAULT when running in WDDM + * + * @note On MIG-enabled GPUs, compute mode would be set to DEFAULT and changing it is not supported. + * + * See \ref nvmlComputeMode_t for details on available compute modes. + * + * @param device The identifier of the target device + * @param mode The target compute mode + * + * @return + * - \ref NVML_SUCCESS if the compute mode was set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetComputeMode() + */ +nvmlReturn_t DECLDIR nvmlDeviceSetComputeMode(nvmlDevice_t device, nvmlComputeMode_t mode); + +/** + * Set the ECC mode for the device. + * + * For Kepler &tm; or newer fully supported devices. + * Only applicable to devices with ECC. + * Requires \a NVML_INFOROM_ECC version 1.0 or higher. + * Requires root/admin permissions. + * + * The ECC mode determines whether the GPU enables its ECC support. + * + * This operation takes effect after the next reboot. + * + * See \ref nvmlEnableState_t for details on available modes. + * + * @param device The identifier of the target device + * @param ecc The target ECC mode + * + * @return + * - \ref NVML_SUCCESS if the ECC mode was set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a ecc is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetEccMode() + */ +nvmlReturn_t DECLDIR nvmlDeviceSetEccMode(nvmlDevice_t device, nvmlEnableState_t ecc); + +/** + * Clear the ECC error and other memory error counts for the device. + * + * For Kepler &tm; or newer fully supported devices. + * Only applicable to devices with ECC. + * Requires \a NVML_INFOROM_ECC version 2.0 or higher to clear aggregate location-based ECC counts. + * Requires \a NVML_INFOROM_ECC version 1.0 or higher to clear all other ECC counts. + * Requires root/admin permissions. + * Requires ECC Mode to be enabled. + * + * Sets all of the specified ECC counters to 0, including both detailed and total counts. + * + * This operation takes effect immediately. + * + * See \ref nvmlMemoryErrorType_t for details on available counter types. + * + * @param device The identifier of the target device + * @param counterType Flag that indicates which type of errors should be cleared. + * + * @return + * - \ref NVML_SUCCESS if the error counts were cleared + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a counterType is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see + * - nvmlDeviceGetDetailedEccErrors() + * - nvmlDeviceGetTotalEccErrors() + */ +nvmlReturn_t DECLDIR nvmlDeviceClearEccErrorCounts(nvmlDevice_t device, nvmlEccCounterType_t counterType); + +/** + * Set the driver model for the device. + * + * For Fermi &tm; or newer fully supported devices. + * For windows only. + * Requires root/admin permissions. + * + * On Windows platforms the device driver can run in either WDDM or WDM (TCC) mode. If a display is attached + * to the device it must run in WDDM mode. + * + * It is possible to force the change to WDM (TCC) while the display is still attached with a force flag (nvmlFlagForce). + * This should only be done if the host is subsequently powered down and the display is detached from the device + * before the next reboot. + * + * This operation takes effect after the next reboot. + * + * Windows driver model may only be set to WDDM when running in DEFAULT compute mode. + * + * Change driver model to WDDM is not supported when GPU doesn't support graphics acceleration or + * will not support it after reboot. See \ref nvmlDeviceSetGpuOperationMode. + * + * See \ref nvmlDriverModel_t for details on available driver models. + * See \ref nvmlFlagDefault and \ref nvmlFlagForce + * + * @param device The identifier of the target device + * @param driverModel The target driver model + * @param flags Flags that change the default behavior + * + * @return + * - \ref NVML_SUCCESS if the driver model has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a driverModel is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the platform is not windows or the device does not support this feature + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlDeviceGetDriverModel() + */ +nvmlReturn_t DECLDIR nvmlDeviceSetDriverModel(nvmlDevice_t device, nvmlDriverModel_t driverModel, unsigned int flags); + +typedef enum nvmlClockLimitId_enum { + NVML_CLOCK_LIMIT_ID_RANGE_START = 0xffffff00, + NVML_CLOCK_LIMIT_ID_TDP, + NVML_CLOCK_LIMIT_ID_UNLIMITED +} nvmlClockLimitId_t; + +/** + * Set clocks that device will lock to. + * + * Sets the clocks that the device will be running at to the value in the range of minGpuClockMHz to maxGpuClockMHz. + * Setting this will supersede application clock values and take effect regardless if a cuda app is running. + * See /ref nvmlDeviceSetApplicationsClocks + * + * Can be used as a setting to request constant performance. + * + * This can be called with a pair of integer clock frequencies in MHz, or a pair of /ref nvmlClockLimitId_t values. + * See the table below for valid combinations of these values. + * + * minGpuClock | maxGpuClock | Effect + * ------------+-------------+-------------------------------------------------- + * tdp | tdp | Lock clock to TDP + * unlimited | tdp | Upper bound is TDP but clock may drift below this + * tdp | unlimited | Lower bound is TDP but clock may boost above this + * unlimited | unlimited | Unlocked (== nvmlDeviceResetGpuLockedClocks) + * + * If one arg takes one of these values, the other must be one of these values as + * well. Mixed numeric and symbolic calls return NVML_ERROR_INVALID_ARGUMENT. + * + * Requires root/admin permissions. + * + * After system reboot or driver reload applications clocks go back to their default value. + * See \ref nvmlDeviceResetGpuLockedClocks. + * + * For Volta &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param minGpuClockMHz Requested minimum gpu clock in MHz + * @param maxGpuClockMHz Requested maximum gpu clock in MHz + * + * @return + * - \ref NVML_SUCCESS if new settings were successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a minGpuClockMHz and \a maxGpuClockMHz + * is not a valid clock combination + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceSetGpuLockedClocks(nvmlDevice_t device, unsigned int minGpuClockMHz, unsigned int maxGpuClockMHz); + +/** + * Resets the gpu clock to the default value + * + * This is the gpu clock that will be used after system reboot or driver reload. + * Default values are idle clocks, but the current values can be changed using \ref nvmlDeviceSetApplicationsClocks. + * + * @see nvmlDeviceSetGpuLockedClocks + * + * For Volta &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * + * @return + * - \ref NVML_SUCCESS if new settings were successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceResetGpuLockedClocks(nvmlDevice_t device); + +/** + * Set memory clocks that device will lock to. + * + * Sets the device's memory clocks to the value in the range of minMemClockMHz to maxMemClockMHz. + * Setting this will supersede application clock values and take effect regardless of whether a cuda app is running. + * See /ref nvmlDeviceSetApplicationsClocks + * + * Can be used as a setting to request constant performance. + * + * Requires root/admin permissions. + * + * After system reboot or driver reload applications clocks go back to their default value. + * See \ref nvmlDeviceResetMemoryLockedClocks. + * + * For Ampere &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param minMemClockMHz Requested minimum memory clock in MHz + * @param maxMemClockMHz Requested maximum memory clock in MHz + * + * @return + * - \ref NVML_SUCCESS if new settings were successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a minGpuClockMHz and \a maxGpuClockMHz + * is not a valid clock combination + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceSetMemoryLockedClocks(nvmlDevice_t device, unsigned int minMemClockMHz, unsigned int maxMemClockMHz); + +/** + * Resets the memory clock to the default value + * + * This is the memory clock that will be used after system reboot or driver reload. + * Default values are idle clocks, but the current values can be changed using \ref nvmlDeviceSetApplicationsClocks. + * + * @see nvmlDeviceSetMemoryLockedClocks + * + * For Ampere &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * + * @return + * - \ref NVML_SUCCESS if new settings were successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceResetMemoryLockedClocks(nvmlDevice_t device); + +/** + * Set clocks that applications will lock to. + * + * Sets the clocks that compute and graphics applications will be running at. + * e.g. CUDA driver requests these clocks during context creation which means this property + * defines clocks at which CUDA applications will be running unless some overspec event + * occurs (e.g. over power, over thermal or external HW brake). + * + * Can be used as a setting to request constant performance. + * + * On Pascal and newer hardware, this will automatically disable automatic boosting of clocks. + * + * On K80 and newer Kepler and Maxwell GPUs, users desiring fixed performance should also call + * \ref nvmlDeviceSetAutoBoostedClocksEnabled to prevent clocks from automatically boosting + * above the clock value being set. + * + * For Kepler &tm; or newer non-GeForce fully supported devices and Maxwell or newer GeForce devices. + * Requires root/admin permissions. + * + * See \ref nvmlDeviceGetSupportedMemoryClocks and \ref nvmlDeviceGetSupportedGraphicsClocks + * for details on how to list available clocks combinations. + * + * After system reboot or driver reload applications clocks go back to their default value. + * See \ref nvmlDeviceResetApplicationsClocks. + * + * @param device The identifier of the target device + * @param memClockMHz Requested memory clock in MHz + * @param graphicsClockMHz Requested graphics clock in MHz + * + * @return + * - \ref NVML_SUCCESS if new settings were successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a memClockMHz and \a graphicsClockMHz + * is not a valid clock combination + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceSetApplicationsClocks(nvmlDevice_t device, unsigned int memClockMHz, unsigned int graphicsClockMHz); + +/** + * Resets the application clock to the default value + * + * This is the applications clock that will be used after system reboot or driver reload. + * Default value is constant, but the current value an be changed using \ref nvmlDeviceSetApplicationsClocks. + * + * On Pascal and newer hardware, if clocks were previously locked with \ref nvmlDeviceSetApplicationsClocks, + * this call will unlock clocks. This returns clocks their default behavior ofautomatically boosting above + * base clocks as thermal limits allow. + * + * @see nvmlDeviceGetApplicationsClock + * @see nvmlDeviceSetApplicationsClocks + * + * For Fermi &tm; or newer non-GeForce fully supported devices and Maxwell or newer GeForce devices. + * + * @param device The identifier of the target device + * + * @return + * - \ref NVML_SUCCESS if new settings were successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceResetApplicationsClocks(nvmlDevice_t device); + /** * Try to set the current state of Auto Boosted clocks on a device. * @@ -2891,61 +6725,57 @@ nvmlReturn_t DECLDIR nvmlDeviceSetAutoBoostedClocksEnabled(nvmlDevice_t device, */ nvmlReturn_t DECLDIR nvmlDeviceSetDefaultAutoBoostedClocksEnabled(nvmlDevice_t device, nvmlEnableState_t enabled, unsigned int flags); - /** - * Retrieves the intended operating speed of the device's fan. + * Sets the speed of the fan control policy to default. * - * Note: The reported speed is the intended fan speed. If the fan is physically blocked and unable to spin, the - * output will not match the actual fan speed. - * - * For all discrete products with dedicated fans. + * For all cuda-capable discrete products with fans * - * The fan speed is expressed as a percent of the maximum, i.e. full speed is 100%. + * @param device The identifier of the target device + * @param fan The index of the fan, starting at zero * - * @param device The identifier of the target device - * @param speed Reference in which to return the fan speed percentage - * - * @return - * - \ref NVML_SUCCESS if \a speed has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a speed is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have a fan - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * return + * NVML_SUCCESS if speed has been adjusted + * NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * NVML_ERROR_INVALID_ARGUMENT if device is invalid + * NVML_ERROR_NOT_SUPPORTED if the device does not support this + * (doesn't have fans) + * NVML_ERROR_UNKNOWN on any unexpected error */ -nvmlReturn_t DECLDIR nvmlDeviceGetFanSpeed(nvmlDevice_t device, unsigned int *speed); +nvmlReturn_t DECLDIR nvmlDeviceSetDefaultFanSpeed_v2(nvmlDevice_t device, unsigned int fan); /** - * Retrieves the current temperature readings for the device, in degrees C. - * - * For all products. + * Sets current fan control policy. * - * See \ref nvmlTemperatureSensors_t for details on available temperature sensors. + * For Maxwell &tm; or newer fully supported devices. * - * @param device The identifier of the target device - * @param sensorType Flag that indicates which sensor reading to retrieve - * @param temp Reference in which to return the temperature reading - * - * @return - * - \ref NVML_SUCCESS if \a temp has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a sensorType is invalid or \a temp is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not have the specified sensor - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * Requires privileged user. + * + * For all cuda-capable discrete products with fans + * + * device The identifier of the target \a device + * policy The fan control \a policy to set + * + * return + * NVML_SUCCESS if \a policy has been set + * NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a policy is null or the \a fan given doesn't reference + * a fan that exists. + * NVML_ERROR_NOT_SUPPORTED if the \a device is older than Maxwell + * NVML_ERROR_UNKNOWN on any unexpected error */ -nvmlReturn_t DECLDIR nvmlDeviceGetTemperature(nvmlDevice_t device, nvmlTemperatureSensors_t sensorType, unsigned int *temp); +nvmlReturn_t DECLDIR nvmlDeviceSetFanControlPolicy(nvmlDevice_t device, unsigned int fan, + nvmlFanControlPolicy_t policy); /** - * Retrieves the temperature threshold for the GPU with the specified threshold type in degrees C. + * Sets the temperature threshold for the GPU with the specified threshold type in degrees C. * - * For Kepler &tm; or newer fully supported devices. + * For Maxwell &tm; or newer fully supported devices. * * See \ref nvmlTemperatureThresholds_t for details on available temperature thresholds. * * @param device The identifier of the target device - * @param thresholdType The type of threshold value queried - * @param temp Reference in which to return the temperature reading + * @param thresholdType The type of threshold value to be set + * @param temp Reference which hold the value to be set * @return * - \ref NVML_SUCCESS if \a temp has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized @@ -2954,1369 +6784,11 @@ nvmlReturn_t DECLDIR nvmlDeviceGetTemperature(nvmlDevice_t device, nvmlTemperatu * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ -nvmlReturn_t DECLDIR nvmlDeviceGetTemperatureThreshold(nvmlDevice_t device, nvmlTemperatureThresholds_t thresholdType, unsigned int *temp); - -/** - * Retrieves the current performance state for the device. - * - * For Fermi &tm; or newer fully supported devices. - * - * See \ref nvmlPstates_t for details on allowed performance states. - * - * @param device The identifier of the target device - * @param pState Reference in which to return the performance state reading - * - * @return - * - \ref NVML_SUCCESS if \a pState has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pState is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetPerformanceState(nvmlDevice_t device, nvmlPstates_t *pState); - -/** - * Retrieves current clocks throttling reasons. - * - * For all fully supported products. - * - * \note More than one bit can be enabled at the same time. Multiple reasons can be affecting clocks at once. - * - * @param device The identifier of the target device - * @param clocksThrottleReasons Reference in which to return bitmask of active clocks throttle - * reasons - * - * @return - * - \ref NVML_SUCCESS if \a clocksThrottleReasons has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a clocksThrottleReasons is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlClocksThrottleReasons - * @see nvmlDeviceGetSupportedClocksThrottleReasons - */ -nvmlReturn_t DECLDIR nvmlDeviceGetCurrentClocksThrottleReasons(nvmlDevice_t device, unsigned long long *clocksThrottleReasons); - -/** - * Retrieves bitmask of supported clocks throttle reasons that can be returned by - * \ref nvmlDeviceGetCurrentClocksThrottleReasons - * - * For all fully supported products. - * - * This method is not supported in virtual machines running virtual GPU (vGPU). - * - * @param device The identifier of the target device - * @param supportedClocksThrottleReasons Reference in which to return bitmask of supported - * clocks throttle reasons - * - * @return - * - \ref NVML_SUCCESS if \a supportedClocksThrottleReasons has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a supportedClocksThrottleReasons is NULL - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlClocksThrottleReasons - * @see nvmlDeviceGetCurrentClocksThrottleReasons - */ -nvmlReturn_t DECLDIR nvmlDeviceGetSupportedClocksThrottleReasons(nvmlDevice_t device, unsigned long long *supportedClocksThrottleReasons); - -/** - * Deprecated: Use \ref nvmlDeviceGetPerformanceState. This function exposes an incorrect generalization. - * - * Retrieve the current performance state for the device. - * - * For Fermi &tm; or newer fully supported devices. - * - * See \ref nvmlPstates_t for details on allowed performance states. - * - * @param device The identifier of the target device - * @param pState Reference in which to return the performance state reading - * - * @return - * - \ref NVML_SUCCESS if \a pState has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pState is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetPowerState(nvmlDevice_t device, nvmlPstates_t *pState); - -/** - * This API has been deprecated. - * - * Retrieves the power management mode associated with this device. - * - * For products from the Fermi family. - * - Requires \a NVML_INFOROM_POWER version 3.0 or higher. - * - * For from the Kepler or newer families. - * - Does not require \a NVML_INFOROM_POWER object. - * - * This flag indicates whether any power management algorithm is currently active on the device. An - * enabled state does not necessarily mean the device is being actively throttled -- only that - * that the driver will do so if the appropriate conditions are met. - * - * See \ref nvmlEnableState_t for details on allowed modes. - * - * @param device The identifier of the target device - * @param mode Reference in which to return the current power management mode - * - * @return - * - \ref NVML_SUCCESS if \a mode has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetPowerManagementMode(nvmlDevice_t device, nvmlEnableState_t *mode); - -/** - * Retrieves the power management limit associated with this device. - * - * For Fermi &tm; or newer fully supported devices. - * - * The power limit defines the upper boundary for the card's power draw. If - * the card's total power draw reaches this limit the power management algorithm kicks in. - * - * This reading is only available if power management mode is supported. - * See \ref nvmlDeviceGetPowerManagementMode. - * - * @param device The identifier of the target device - * @param limit Reference in which to return the power management limit in milliwatts - * - * @return - * - \ref NVML_SUCCESS if \a limit has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a limit is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetPowerManagementLimit(nvmlDevice_t device, unsigned int *limit); - -/** - * Retrieves information about possible values of power management limits on this device. - * - * For Kepler &tm; or newer fully supported devices. - * - * @param device The identifier of the target device - * @param minLimit Reference in which to return the minimum power management limit in milliwatts - * @param maxLimit Reference in which to return the maximum power management limit in milliwatts - * - * @return - * - \ref NVML_SUCCESS if \a minLimit and \a maxLimit have been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a minLimit or \a maxLimit is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlDeviceSetPowerManagementLimit - */ -nvmlReturn_t DECLDIR nvmlDeviceGetPowerManagementLimitConstraints(nvmlDevice_t device, unsigned int *minLimit, unsigned int *maxLimit); - -/** - * Retrieves default power management limit on this device, in milliwatts. - * Default power management limit is a power management limit that the device boots with. - * - * For Kepler &tm; or newer fully supported devices. - * - * @param device The identifier of the target device - * @param defaultLimit Reference in which to return the default power management limit in milliwatts - * - * @return - * - \ref NVML_SUCCESS if \a defaultLimit has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a defaultLimit is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetPowerManagementDefaultLimit(nvmlDevice_t device, unsigned int *defaultLimit); - -/** - * Retrieves power usage for this GPU in milliwatts and its associated circuitry (e.g. memory) - * - * For Fermi &tm; or newer fully supported devices. - * - * On Fermi and Kepler GPUs the reading is accurate to within +/- 5% of current power draw. - * - * It is only available if power management mode is supported. See \ref nvmlDeviceGetPowerManagementMode. - * - * @param device The identifier of the target device - * @param power Reference in which to return the power usage information - * - * @return - * - \ref NVML_SUCCESS if \a power has been populated - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a power is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support power readings - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetPowerUsage(nvmlDevice_t device, unsigned int *power); - -/** - * Retrieves total energy consumption for this GPU in millijoules (mJ) since the driver was last reloaded - * - * For newer than Pascal &tm; fully supported devices. - * - * @param device The identifier of the target device - * @param energy Reference in which to return the energy consumption information - * - * @return - * - \ref NVML_SUCCESS if \a energy has been populated - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a energy is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support energy readings - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetTotalEnergyConsumption(nvmlDevice_t device, unsigned long long *energy); - -/** - * Get the effective power limit that the driver enforces after taking into account all limiters - * - * Note: This can be different from the \ref nvmlDeviceGetPowerManagementLimit if other limits are set elsewhere - * This includes the out of band power limit interface - * - * For Kepler &tm; or newer fully supported devices. - * - * @param device The device to communicate with - * @param limit Reference in which to return the power management limit in milliwatts - * - * @return - * - \ref NVML_SUCCESS if \a limit has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a limit is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetEnforcedPowerLimit(nvmlDevice_t device, unsigned int *limit); - -/** - * Retrieves the current GOM and pending GOM (the one that GPU will switch to after reboot). - * - * For GK110 M-class and X-class Tesla &tm; products from the Kepler family. - * Modes \ref NVML_GOM_LOW_DP and \ref NVML_GOM_ALL_ON are supported on fully supported GeForce products. - * Not supported on Quadro ® and Tesla &tm; C-class products. - * - * @param device The identifier of the target device - * @param current Reference in which to return the current GOM - * @param pending Reference in which to return the pending GOM - * - * @return - * - \ref NVML_SUCCESS if \a mode has been populated - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a current or \a pending is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlGpuOperationMode_t - * @see nvmlDeviceSetGpuOperationMode - */ -nvmlReturn_t DECLDIR nvmlDeviceGetGpuOperationMode(nvmlDevice_t device, nvmlGpuOperationMode_t *current, nvmlGpuOperationMode_t *pending); - -/** - * Retrieves the amount of used, free and total memory available on the device, in bytes. - * - * For all products. - * - * Enabling ECC reduces the amount of total available memory, due to the extra required parity bits. - * Under WDDM most device memory is allocated and managed on startup by Windows. - * - * Under Linux and Windows TCC, the reported amount of used memory is equal to the sum of memory allocated - * by all active channels on the device. - * - * See \ref nvmlMemory_t for details on available memory info. - * - * @param device The identifier of the target device - * @param memory Reference in which to return the memory information - * - * @return - * - \ref NVML_SUCCESS if \a memory has been populated - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a memory is NULL - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetMemoryInfo(nvmlDevice_t device, nvmlMemory_t *memory); - -/** - * Retrieves the current compute mode for the device. - * - * For all products. - * - * See \ref nvmlComputeMode_t for details on allowed compute modes. - * - * @param device The identifier of the target device - * @param mode Reference in which to return the current compute mode - * - * @return - * - \ref NVML_SUCCESS if \a mode has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlDeviceSetComputeMode() - */ -nvmlReturn_t DECLDIR nvmlDeviceGetComputeMode(nvmlDevice_t device, nvmlComputeMode_t *mode); - -/** - * Retrieves the CUDA compute capability of the device. - * - * For all products. - * - * Returns the major and minor compute capability version numbers of the - * device. The major and minor versions are equivalent to the - * CU_DEVICE_ATTRIBUTE_COMPUTE_CAPABILITY_MINOR and - * CU_DEVICE_ATTRIBUTE_COMPUTE_CAPABILITY_MAJOR attributes that would be - * returned by CUDA's cuDeviceGetAttribute(). - * - * @param device The identifier of the target device - * @param major Reference in which to return the major CUDA compute capability - * @param minor Reference in which to return the minor CUDA compute capability - * - * @return - * - \ref NVML_SUCCESS if \a major and \a minor have been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a major or \a minor are NULL - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetCudaComputeCapability(nvmlDevice_t device, int *major, int *minor); - -/** - * Retrieves the current and pending ECC modes for the device. - * - * For Fermi &tm; or newer fully supported devices. - * Only applicable to devices with ECC. - * Requires \a NVML_INFOROM_ECC version 1.0 or higher. - * - * Changing ECC modes requires a reboot. The "pending" ECC mode refers to the target mode following - * the next reboot. - * - * See \ref nvmlEnableState_t for details on allowed modes. - * - * @param device The identifier of the target device - * @param current Reference in which to return the current ECC mode - * @param pending Reference in which to return the pending ECC mode - * - * @return - * - \ref NVML_SUCCESS if \a current and \a pending have been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or either \a current or \a pending is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlDeviceSetEccMode() - */ -nvmlReturn_t DECLDIR nvmlDeviceGetEccMode(nvmlDevice_t device, nvmlEnableState_t *current, nvmlEnableState_t *pending); - -/** - * Retrieves the device boardId from 0-N. - * Devices with the same boardId indicate GPUs connected to the same PLX. Use in conjunction with - * \ref nvmlDeviceGetMultiGpuBoard() to decide if they are on the same board as well. - * The boardId returned is a unique ID for the current configuration. Uniqueness and ordering across - * reboots and system configurations is not guaranteed (i.e. if a Tesla K40c returns 0x100 and - * the two GPUs on a Tesla K10 in the same system returns 0x200 it is not guaranteed they will - * always return those values but they will always be different from each other). - * - * - * For Fermi &tm; or newer fully supported devices. - * - * @param device The identifier of the target device - * @param boardId Reference in which to return the device's board ID - * - * @return - * - \ref NVML_SUCCESS if \a boardId has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a boardId is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetBoardId(nvmlDevice_t device, unsigned int *boardId); - -/** - * Retrieves whether the device is on a Multi-GPU Board - * Devices that are on multi-GPU boards will set \a multiGpuBool to a non-zero value. - * - * For Fermi &tm; or newer fully supported devices. - * - * @param device The identifier of the target device - * @param multiGpuBool Reference in which to return a zero or non-zero value - * to indicate whether the device is on a multi GPU board - * - * @return - * - \ref NVML_SUCCESS if \a multiGpuBool has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a multiGpuBool is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetMultiGpuBoard(nvmlDevice_t device, unsigned int *multiGpuBool); - -/** - * Retrieves the total ECC error counts for the device. - * - * For Fermi &tm; or newer fully supported devices. - * Only applicable to devices with ECC. - * Requires \a NVML_INFOROM_ECC version 1.0 or higher. - * Requires ECC Mode to be enabled. - * - * The total error count is the sum of errors across each of the separate memory systems, i.e. the total set of - * errors across the entire device. - * - * See \ref nvmlMemoryErrorType_t for a description of available error types.\n - * See \ref nvmlEccCounterType_t for a description of available counter types. - * - * @param device The identifier of the target device - * @param errorType Flag that specifies the type of the errors. - * @param counterType Flag that specifies the counter-type of the errors. - * @param eccCounts Reference in which to return the specified ECC errors - * - * @return - * - \ref NVML_SUCCESS if \a eccCounts has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a errorType or \a counterType is invalid, or \a eccCounts is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlDeviceClearEccErrorCounts() - */ -nvmlReturn_t DECLDIR nvmlDeviceGetTotalEccErrors(nvmlDevice_t device, nvmlMemoryErrorType_t errorType, nvmlEccCounterType_t counterType, unsigned long long *eccCounts); - -/** - * Retrieves the detailed ECC error counts for the device. - * - * @deprecated This API supports only a fixed set of ECC error locations - * On different GPU architectures different locations are supported - * See \ref nvmlDeviceGetMemoryErrorCounter - * - * For Fermi &tm; or newer fully supported devices. - * Only applicable to devices with ECC. - * Requires \a NVML_INFOROM_ECC version 2.0 or higher to report aggregate location-based ECC counts. - * Requires \a NVML_INFOROM_ECC version 1.0 or higher to report all other ECC counts. - * Requires ECC Mode to be enabled. - * - * Detailed errors provide separate ECC counts for specific parts of the memory system. - * - * Reports zero for unsupported ECC error counters when a subset of ECC error counters are supported. - * - * See \ref nvmlMemoryErrorType_t for a description of available bit types.\n - * See \ref nvmlEccCounterType_t for a description of available counter types.\n - * See \ref nvmlEccErrorCounts_t for a description of provided detailed ECC counts. - * - * @param device The identifier of the target device - * @param errorType Flag that specifies the type of the errors. - * @param counterType Flag that specifies the counter-type of the errors. - * @param eccCounts Reference in which to return the specified ECC errors - * - * @return - * - \ref NVML_SUCCESS if \a eccCounts has been populated - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a errorType or \a counterType is invalid, or \a eccCounts is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlDeviceClearEccErrorCounts() - */ -nvmlReturn_t DECLDIR nvmlDeviceGetDetailedEccErrors(nvmlDevice_t device, nvmlMemoryErrorType_t errorType, nvmlEccCounterType_t counterType, nvmlEccErrorCounts_t *eccCounts); - -/** - * Retrieves the requested memory error counter for the device. - * - * For Fermi &tm; or newer fully supported devices. - * Requires \a NVML_INFOROM_ECC version 2.0 or higher to report aggregate location-based memory error counts. - * Requires \a NVML_INFOROM_ECC version 1.0 or higher to report all other memory error counts. - * - * Only applicable to devices with ECC. - * - * Requires ECC Mode to be enabled. - * - * See \ref nvmlMemoryErrorType_t for a description of available memory error types.\n - * See \ref nvmlEccCounterType_t for a description of available counter types.\n - * See \ref nvmlMemoryLocation_t for a description of available counter locations.\n - * - * @param device The identifier of the target device - * @param errorType Flag that specifies the type of error. - * @param counterType Flag that specifies the counter-type of the errors. - * @param locationType Specifies the location of the counter. - * @param count Reference in which to return the ECC counter - * - * @return - * - \ref NVML_SUCCESS if \a count has been populated - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a bitTyp,e \a counterType or \a locationType is - * invalid, or \a count is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support ECC error reporting in the specified memory - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetMemoryErrorCounter(nvmlDevice_t device, nvmlMemoryErrorType_t errorType, - nvmlEccCounterType_t counterType, - nvmlMemoryLocation_t locationType, unsigned long long *count); - -/** - * Retrieves the current utilization rates for the device's major subsystems. - * - * For Fermi &tm; or newer fully supported devices. - * - * See \ref nvmlUtilization_t for details on available utilization rates. - * - * \note During driver initialization when ECC is enabled one can see high GPU and Memory Utilization readings. - * This is caused by ECC Memory Scrubbing mechanism that is performed during driver initialization. - * - * @param device The identifier of the target device - * @param utilization Reference in which to return the utilization information - * - * @return - * - \ref NVML_SUCCESS if \a utilization has been populated - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a utilization is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetUtilizationRates(nvmlDevice_t device, nvmlUtilization_t *utilization); - -/** - * Retrieves the current utilization and sampling size in microseconds for the Encoder - * - * For Kepler &tm; or newer fully supported devices. - * - * @param device The identifier of the target device - * @param utilization Reference to an unsigned int for encoder utilization info - * @param samplingPeriodUs Reference to an unsigned int for the sampling period in US - * - * @return - * - \ref NVML_SUCCESS if \a utilization has been populated - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a utilization is NULL, or \a samplingPeriodUs is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetEncoderUtilization(nvmlDevice_t device, unsigned int *utilization, unsigned int *samplingPeriodUs); - -/** - * Retrieves the current capacity of the device's encoder, in macroblocks per second. - * - * For Maxwell &tm; or newer fully supported devices. - * - * @param device The identifier of the target device - * @param encoderQueryType Type of encoder to query - * @param encoderCapacity Reference to an unsigned int for the encoder capacity - * - * @return - * - \ref NVML_SUCCESS if \a encoderCapacity is fetched - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a encoderCapacity is NULL, or \a device or \a encoderQueryType - * are invalid - * - \ref NVML_ERROR_NOT_SUPPORTED if device does not support the encoder specified in \a encodeQueryType - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetEncoderCapacity (nvmlDevice_t device, nvmlEncoderType_t encoderQueryType, unsigned int *encoderCapacity); - -/** - * Retrieves the current encoder statistics for a given device. - * - * For Maxwell &tm; or newer fully supported devices. - * - * @param device The identifier of the target device - * @param sessionCount Reference to an unsigned int for count of active encoder sessions - * @param averageFps Reference to an unsigned int for trailing average FPS of all active sessions - * @param averageLatency Reference to an unsigned int for encode latency in microseconds - * - * @return - * - \ref NVML_SUCCESS if \a sessionCount, \a averageFps and \a averageLatency is fetched - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a sessionCount, or \a device or \a averageFps, - * or \a averageLatency is NULL - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetEncoderStats (nvmlDevice_t device, unsigned int *sessionCount, - unsigned int *averageFps, unsigned int *averageLatency); - -/** - * Retrieves information about active encoder sessions on a target device. - * - * An array of active encoder sessions is returned in the caller-supplied buffer pointed at by \a sessionInfos. The - * array elememt count is passed in \a sessionCount, and \a sessionCount is used to return the number of sessions - * written to the buffer. - * - * If the supplied buffer is not large enough to accommodate the active session array, the function returns - * NVML_ERROR_INSUFFICIENT_SIZE, with the element count of nvmlEncoderSessionInfo_t array required in \a sessionCount. - * To query the number of active encoder sessions, call this function with *sessionCount = 0. The code will return - * NVML_SUCCESS with number of active encoder sessions updated in *sessionCount. - * - * For Maxwell &tm; or newer fully supported devices. - * - * @param device The identifier of the target device - * @param sessionCount Reference to caller supplied array size, and returns the number of sessions. - * @param sessionInfos Reference in which to return the session information - * - * @return - * - \ref NVML_SUCCESS if \a sessionInfos is fetched - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a sessionCount is too small, array element count is returned in \a sessionCount - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a sessionCount is NULL. - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetEncoderSessions(nvmlDevice_t device, unsigned int *sessionCount, nvmlEncoderSessionInfo_t *sessionInfos); - -/** - * Retrieves the current utilization and sampling size in microseconds for the Decoder - * - * For Kepler &tm; or newer fully supported devices. - * - * @param device The identifier of the target device - * @param utilization Reference to an unsigned int for decoder utilization info - * @param samplingPeriodUs Reference to an unsigned int for the sampling period in US - * - * @return - * - \ref NVML_SUCCESS if \a utilization has been populated - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a utilization is NULL, or \a samplingPeriodUs is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetDecoderUtilization(nvmlDevice_t device, unsigned int *utilization, unsigned int *samplingPeriodUs); - -/** - * Retrieves the current and pending driver model for the device. - * - * For Fermi &tm; or newer fully supported devices. - * For windows only. - * - * On Windows platforms the device driver can run in either WDDM or WDM (TCC) mode. If a display is attached - * to the device it must run in WDDM mode. TCC mode is preferred if a display is not attached. - * - * See \ref nvmlDriverModel_t for details on available driver models. - * - * @param device The identifier of the target device - * @param current Reference in which to return the current driver model - * @param pending Reference in which to return the pending driver model - * - * @return - * - \ref NVML_SUCCESS if either \a current and/or \a pending have been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or both \a current and \a pending are NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the platform is not windows - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlDeviceSetDriverModel() - */ -nvmlReturn_t DECLDIR nvmlDeviceGetDriverModel(nvmlDevice_t device, nvmlDriverModel_t *current, nvmlDriverModel_t *pending); - -/** - * Get VBIOS version of the device. - * - * For all products. - * - * The VBIOS version may change from time to time. It will not exceed 32 characters in length - * (including the NULL terminator). See \ref nvmlConstants::NVML_DEVICE_VBIOS_VERSION_BUFFER_SIZE. - * - * @param device The identifier of the target device - * @param version Reference to which to return the VBIOS version - * @param length The maximum allowed length of the string returned in \a version - * - * @return - * - \ref NVML_SUCCESS if \a version has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a version is NULL - * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetVbiosVersion(nvmlDevice_t device, char *version, unsigned int length); - -/** - * Get Bridge Chip Information for all the bridge chips on the board. - * - * For all fully supported products. - * Only applicable to multi-GPU products. - * - * @param device The identifier of the target device - * @param bridgeHierarchy Reference to the returned bridge chip Hierarchy - * - * @return - * - \ref NVML_SUCCESS if bridge chip exists - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a bridgeInfo is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if bridge chip not supported on the device - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - */ -nvmlReturn_t DECLDIR nvmlDeviceGetBridgeChipInfo(nvmlDevice_t device, nvmlBridgeChipHierarchy_t *bridgeHierarchy); - -/** - * Get information about processes with a compute context on a device - * - * For Fermi &tm; or newer fully supported devices. - * - * This function returns information only about compute running processes (e.g. CUDA application which have - * active context). Any graphics applications (e.g. using OpenGL, DirectX) won't be listed by this function. - * - * To query the current number of running compute processes, call this function with *infoCount = 0. The - * return code will be NVML_ERROR_INSUFFICIENT_SIZE, or NVML_SUCCESS if none are running. For this call - * \a infos is allowed to be NULL. - * - * The usedGpuMemory field returned is all of the memory used by the application. - * - * Keep in mind that information returned by this call is dynamic and the number of elements might change in - * time. Allocate more space for \a infos table in case new compute processes are spawned. - * - * @param device The identifier of the target device - * @param infoCount Reference in which to provide the \a infos array size, and - * to return the number of returned elements - * @param infos Reference in which to return the process information - * - * @return - * - \ref NVML_SUCCESS if \a infoCount and \a infos have been populated - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a infoCount indicates that the \a infos array is too small - * \a infoCount will contain minimal amount of space necessary for - * the call to complete - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, either of \a infoCount or \a infos is NULL - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see \ref nvmlSystemGetProcessName - */ -nvmlReturn_t DECLDIR nvmlDeviceGetComputeRunningProcesses(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_t *infos); - -/** - * Get information about processes with a graphics context on a device - * - * For Kepler &tm; or newer fully supported devices. - * - * This function returns information only about graphics based processes - * (eg. applications using OpenGL, DirectX) - * - * To query the current number of running graphics processes, call this function with *infoCount = 0. The - * return code will be NVML_ERROR_INSUFFICIENT_SIZE, or NVML_SUCCESS if none are running. For this call - * \a infos is allowed to be NULL. - * - * The usedGpuMemory field returned is all of the memory used by the application. - * - * Keep in mind that information returned by this call is dynamic and the number of elements might change in - * time. Allocate more space for \a infos table in case new graphics processes are spawned. - * - * @param device The identifier of the target device - * @param infoCount Reference in which to provide the \a infos array size, and - * to return the number of returned elements - * @param infos Reference in which to return the process information - * - * @return - * - \ref NVML_SUCCESS if \a infoCount and \a infos have been populated - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a infoCount indicates that the \a infos array is too small - * \a infoCount will contain minimal amount of space necessary for - * the call to complete - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, either of \a infoCount or \a infos is NULL - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see \ref nvmlSystemGetProcessName - */ -nvmlReturn_t DECLDIR nvmlDeviceGetGraphicsRunningProcesses(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_t *infos); - -/** - * Check if the GPU devices are on the same physical board. - * - * For all fully supported products. - * - * @param device1 The first GPU device - * @param device2 The second GPU device - * @param onSameBoard Reference in which to return the status. - * Non-zero indicates that the GPUs are on the same board. - * - * @return - * - \ref NVML_SUCCESS if \a onSameBoard has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a dev1 or \a dev2 are invalid or \a onSameBoard is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if this check is not supported by the device - * - \ref NVML_ERROR_GPU_IS_LOST if the either GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceOnSameBoard(nvmlDevice_t device1, nvmlDevice_t device2, int *onSameBoard); - -/** - * Retrieves the root/admin permissions on the target API. See \a nvmlRestrictedAPI_t for the list of supported APIs. - * If an API is restricted only root users can call that API. See \a nvmlDeviceSetAPIRestriction to change current permissions. - * - * For all fully supported products. - * - * @param device The identifier of the target device - * @param apiType Target API type for this operation - * @param isRestricted Reference in which to return the current restriction - * NVML_FEATURE_ENABLED indicates that the API is root-only - * NVML_FEATURE_DISABLED indicates that the API is accessible to all users - * - * @return - * - \ref NVML_SUCCESS if \a isRestricted has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a apiType incorrect or \a isRestricted is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device or the device does not support - * the feature that is being queried (E.G. Enabling/disabling Auto Boosted clocks is - * not supported by the device) - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlRestrictedAPI_t - */ -nvmlReturn_t DECLDIR nvmlDeviceGetAPIRestriction(nvmlDevice_t device, nvmlRestrictedAPI_t apiType, nvmlEnableState_t *isRestricted); - -/** - * Gets recent samples for the GPU. - * - * For Kepler &tm; or newer fully supported devices. - * - * Based on type, this method can be used to fetch the power, utilization or clock samples maintained in the buffer by - * the driver. - * - * Power, Utilization and Clock samples are returned as type "unsigned int" for the union nvmlValue_t. - * - * To get the size of samples that user needs to allocate, the method is invoked with samples set to NULL. - * The returned samplesCount will provide the number of samples that can be queried. The user needs to - * allocate the buffer with size as samplesCount * sizeof(nvmlSample_t). - * - * lastSeenTimeStamp represents CPU timestamp in microseconds. Set it to 0 to fetch all the samples maintained by the - * underlying buffer. Set lastSeenTimeStamp to one of the timeStamps retrieved from the date of the previous query - * to get more recent samples. - * - * This method fetches the number of entries which can be accommodated in the provided samples array, and the - * reference samplesCount is updated to indicate how many samples were actually retrieved. The advantage of using this - * method for samples in contrast to polling via existing methods is to get get higher frequency data at lower polling cost. - * - * @param device The identifier for the target device - * @param type Type of sampling event - * @param lastSeenTimeStamp Return only samples with timestamp greater than lastSeenTimeStamp. - * @param sampleValType Output parameter to represent the type of sample value as described in nvmlSampleVal_t - * @param sampleCount Reference to provide the number of elements which can be queried in samples array - * @param samples Reference in which samples are returned - - * @return - * - \ref NVML_SUCCESS if samples are successfully retrieved - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a samplesCount is NULL or - * reference to \a sampleCount is 0 for non null \a samples - * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_NOT_FOUND if sample entries are not found - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetSamples(nvmlDevice_t device, nvmlSamplingType_t type, unsigned long long lastSeenTimeStamp, - nvmlValueType_t *sampleValType, unsigned int *sampleCount, nvmlSample_t *samples); - -/** - * Gets Total, Available and Used size of BAR1 memory. - * - * BAR1 is used to map the FB (device memory) so that it can be directly accessed by the CPU or by 3rd party - * devices (peer-to-peer on the PCIE bus). - * - * For Kepler &tm; or newer fully supported devices. - * - * @param device The identifier of the target device - * @param bar1Memory Reference in which BAR1 memory - * information is returned. - * - * @return - * - \ref NVML_SUCCESS if BAR1 memory is successfully retrieved - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a bar1Memory is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - */ -nvmlReturn_t DECLDIR nvmlDeviceGetBAR1MemoryInfo(nvmlDevice_t device, nvmlBAR1Memory_t *bar1Memory); - - -/** - * Gets the duration of time during which the device was throttled (lower than requested clocks) due to power - * or thermal constraints. - * - * The method is important to users who are tying to understand if their GPUs throttle at any point during their applications. The - * difference in violation times at two different reference times gives the indication of GPU throttling event. - * - * Violation for thermal capping is not supported at this time. - * - * For Kepler &tm; or newer fully supported devices. - * - * @param device The identifier of the target device - * @param perfPolicyType Represents Performance policy which can trigger GPU throttling - * @param violTime Reference to which violation time related information is returned - * - * - * @return - * - \ref NVML_SUCCESS if violation time is successfully retrieved - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a perfPolicyType is invalid, or \a violTime is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - */ -nvmlReturn_t DECLDIR nvmlDeviceGetViolationStatus(nvmlDevice_t device, nvmlPerfPolicyType_t perfPolicyType, nvmlViolationTime_t *violTime); - -/** - * @} - */ - -/** @addtogroup nvmlAccountingStats - * @{ - */ - -/** - * Queries the state of per process accounting mode. - * - * For Kepler &tm; or newer fully supported devices. - * - * See \ref nvmlDeviceGetAccountingStats for more details. - * See \ref nvmlDeviceSetAccountingMode - * - * @param device The identifier of the target device - * @param mode Reference in which to return the current accounting mode - * - * @return - * - \ref NVML_SUCCESS if the mode has been successfully retrieved - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode are NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetAccountingMode(nvmlDevice_t device, nvmlEnableState_t *mode); - -/** - * Queries process's accounting stats. - * - * For Kepler &tm; or newer fully supported devices. - * - * Accounting stats capture GPU utilization and other statistics across the lifetime of a process. - * Accounting stats can be queried during life time of the process and after its termination. - * The time field in \ref nvmlAccountingStats_t is reported as 0 during the lifetime of the process and - * updated to actual running time after its termination. - * Accounting stats are kept in a circular buffer, newly created processes overwrite information about old - * processes. - * - * See \ref nvmlAccountingStats_t for description of each returned metric. - * List of processes that can be queried can be retrieved from \ref nvmlDeviceGetAccountingPids. - * - * @note Accounting Mode needs to be on. See \ref nvmlDeviceGetAccountingMode. - * @note Only compute and graphics applications stats can be queried. Monitoring applications stats can't be - * queried since they don't contribute to GPU utilization. - * @note In case of pid collision stats of only the latest process (that terminated last) will be reported - * - * @warning On Kepler devices per process statistics are accurate only if there's one process running on a GPU. - * - * @param device The identifier of the target device - * @param pid Process Id of the target process to query stats for - * @param stats Reference in which to return the process's accounting stats - * - * @return - * - \ref NVML_SUCCESS if stats have been successfully retrieved - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a stats are NULL - * - \ref NVML_ERROR_NOT_FOUND if process stats were not found - * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature or accounting mode is disabled - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlDeviceGetAccountingBufferSize - */ -nvmlReturn_t DECLDIR nvmlDeviceGetAccountingStats(nvmlDevice_t device, unsigned int pid, nvmlAccountingStats_t *stats); - -/** - * Queries list of processes that can be queried for accounting stats. The list of processes returned - * can be in running or terminated state. - * - * For Kepler &tm; or newer fully supported devices. - * - * To just query the number of processes ready to be queried, call this function with *count = 0 and - * pids=NULL. The return code will be NVML_ERROR_INSUFFICIENT_SIZE, or NVML_SUCCESS if list is empty. - * - * For more details see \ref nvmlDeviceGetAccountingStats. - * - * @note In case of PID collision some processes might not be accessible before the circular buffer is full. - * - * @param device The identifier of the target device - * @param count Reference in which to provide the \a pids array size, and - * to return the number of elements ready to be queried - * @param pids Reference in which to return list of process ids - * - * @return - * - \ref NVML_SUCCESS if pids were successfully retrieved - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a count is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature or accounting mode is disabled - * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a count is too small (\a count is set to - * expected value) - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlDeviceGetAccountingBufferSize - */ -nvmlReturn_t DECLDIR nvmlDeviceGetAccountingPids(nvmlDevice_t device, unsigned int *count, unsigned int *pids); - -/** - * Returns the number of processes that the circular buffer with accounting pids can hold. - * - * For Kepler &tm; or newer fully supported devices. - * - * This is the maximum number of processes that accounting information will be stored for before information - * about oldest processes will get overwritten by information about new processes. - * - * @param device The identifier of the target device - * @param bufferSize Reference in which to provide the size (in number of elements) - * of the circular buffer for accounting stats. - * - * @return - * - \ref NVML_SUCCESS if buffer size was successfully retrieved - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a bufferSize is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature or accounting mode is disabled - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlDeviceGetAccountingStats - * @see nvmlDeviceGetAccountingPids - */ -nvmlReturn_t DECLDIR nvmlDeviceGetAccountingBufferSize(nvmlDevice_t device, unsigned int *bufferSize); - -/** @} */ - -/** @addtogroup nvmlDeviceQueries - * @{ - */ - -/** - * Returns the list of retired pages by source, including pages that are pending retirement - * The address information provided from this API is the hardware address of the page that was retired. Note - * that this does not match the virtual address used in CUDA, but will match the address information in XID 63 - * - * For Kepler &tm; or newer fully supported devices. - * - * @param device The identifier of the target device - * @param cause Filter page addresses by cause of retirement - * @param pageCount Reference in which to provide the \a addresses buffer size, and - * to return the number of retired pages that match \a cause - * Set to 0 to query the size without allocating an \a addresses buffer - * @param addresses Buffer to write the page addresses into - * - * @return - * - \ref NVML_SUCCESS if \a pageCount was populated and \a addresses was filled - * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a pageCount indicates the buffer is not large enough to store all the - * matching page addresses. \a pageCount is set to the needed size. - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a pageCount is NULL, \a cause is invalid, or - * \a addresses is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetRetiredPages(nvmlDevice_t device, nvmlPageRetirementCause_t cause, - unsigned int *pageCount, unsigned long long *addresses); - -/** - * Check if any pages are pending retirement and need a reboot to fully retire. - * - * For Kepler &tm; or newer fully supported devices. - * - * @param device The identifier of the target device - * @param isPending Reference in which to return the pending status - * - * @return - * - \ref NVML_SUCCESS if \a isPending was populated - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a isPending is NULL - * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceGetRetiredPagesPendingStatus(nvmlDevice_t device, nvmlEnableState_t *isPending); - -/** @} */ - -/***************************************************************************************************/ -/** @defgroup nvmlUnitCommands Unit Commands - * This chapter describes NVML operations that change the state of the unit. For S-class products. - * Each of these requires root/admin access. Non-admin users will see an NVML_ERROR_NO_PERMISSION - * error code when invoking any of these methods. - * @{ - */ -/***************************************************************************************************/ - -/** - * Set the LED state for the unit. The LED can be either green (0) or amber (1). - * - * For S-class products. - * Requires root/admin permissions. - * - * This operation takes effect immediately. - * - * - * Current S-Class products don't provide unique LEDs for each unit. As such, both front - * and back LEDs will be toggled in unison regardless of which unit is specified with this command. - * - * See \ref nvmlLedColor_t for available colors. - * - * @param unit The identifier of the target unit - * @param color The target LED color - * - * @return - * - \ref NVML_SUCCESS if the LED color has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a unit or \a color is invalid - * - \ref NVML_ERROR_NOT_SUPPORTED if this is not an S-class product - * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlUnitGetLedState() - */ -nvmlReturn_t DECLDIR nvmlUnitSetLedState(nvmlUnit_t unit, nvmlLedColor_t color); - -/** @} */ - -/***************************************************************************************************/ -/** @defgroup nvmlDeviceCommands Device Commands - * This chapter describes NVML operations that change the state of the device. - * Each of these requires root/admin access. Non-admin users will see an NVML_ERROR_NO_PERMISSION - * error code when invoking any of these methods. - * @{ - */ -/***************************************************************************************************/ - -/** - * Set the persistence mode for the device. - * - * For all products. - * For Linux only. - * Requires root/admin permissions. - * - * The persistence mode determines whether the GPU driver software is torn down after the last client - * exits. - * - * This operation takes effect immediately. It is not persistent across reboots. After each reboot the - * persistence mode is reset to "Disabled". - * - * See \ref nvmlEnableState_t for available modes. - * - * @param device The identifier of the target device - * @param mode The target persistence mode - * - * @return - * - \ref NVML_SUCCESS if the persistence mode was set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode is invalid - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlDeviceGetPersistenceMode() - */ -nvmlReturn_t DECLDIR nvmlDeviceSetPersistenceMode(nvmlDevice_t device, nvmlEnableState_t mode); - -/** - * Set the compute mode for the device. - * - * For all products. - * Requires root/admin permissions. - * - * The compute mode determines whether a GPU can be used for compute operations and whether it can - * be shared across contexts. - * - * This operation takes effect immediately. Under Linux it is not persistent across reboots and - * always resets to "Default". Under windows it is persistent. - * - * Under windows compute mode may only be set to DEFAULT when running in WDDM - * - * See \ref nvmlComputeMode_t for details on available compute modes. - * - * @param device The identifier of the target device - * @param mode The target compute mode - * - * @return - * - \ref NVML_SUCCESS if the compute mode was set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode is invalid - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlDeviceGetComputeMode() - */ -nvmlReturn_t DECLDIR nvmlDeviceSetComputeMode(nvmlDevice_t device, nvmlComputeMode_t mode); - -/** - * Set the ECC mode for the device. - * - * For Kepler &tm; or newer fully supported devices. - * Only applicable to devices with ECC. - * Requires \a NVML_INFOROM_ECC version 1.0 or higher. - * Requires root/admin permissions. - * - * The ECC mode determines whether the GPU enables its ECC support. - * - * This operation takes effect after the next reboot. - * - * See \ref nvmlEnableState_t for details on available modes. - * - * @param device The identifier of the target device - * @param ecc The target ECC mode - * - * @return - * - \ref NVML_SUCCESS if the ECC mode was set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a ecc is invalid - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlDeviceGetEccMode() - */ -nvmlReturn_t DECLDIR nvmlDeviceSetEccMode(nvmlDevice_t device, nvmlEnableState_t ecc); - -/** - * Clear the ECC error and other memory error counts for the device. - * - * For Kepler &tm; or newer fully supported devices. - * Only applicable to devices with ECC. - * Requires \a NVML_INFOROM_ECC version 2.0 or higher to clear aggregate location-based ECC counts. - * Requires \a NVML_INFOROM_ECC version 1.0 or higher to clear all other ECC counts. - * Requires root/admin permissions. - * Requires ECC Mode to be enabled. - * - * Sets all of the specified ECC counters to 0, including both detailed and total counts. - * - * This operation takes effect immediately. - * - * See \ref nvmlMemoryErrorType_t for details on available counter types. - * - * @param device The identifier of the target device - * @param counterType Flag that indicates which type of errors should be cleared. - * - * @return - * - \ref NVML_SUCCESS if the error counts were cleared - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a counterType is invalid - * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature - * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see - * - nvmlDeviceGetDetailedEccErrors() - * - nvmlDeviceGetTotalEccErrors() - */ -nvmlReturn_t DECLDIR nvmlDeviceClearEccErrorCounts(nvmlDevice_t device, nvmlEccCounterType_t counterType); - -/** - * Set the driver model for the device. - * - * For Fermi &tm; or newer fully supported devices. - * For windows only. - * Requires root/admin permissions. - * - * On Windows platforms the device driver can run in either WDDM or WDM (TCC) mode. If a display is attached - * to the device it must run in WDDM mode. - * - * It is possible to force the change to WDM (TCC) while the display is still attached with a force flag (nvmlFlagForce). - * This should only be done if the host is subsequently powered down and the display is detached from the device - * before the next reboot. - * - * This operation takes effect after the next reboot. - * - * Windows driver model may only be set to WDDM when running in DEFAULT compute mode. - * - * Change driver model to WDDM is not supported when GPU doesn't support graphics acceleration or - * will not support it after reboot. See \ref nvmlDeviceSetGpuOperationMode. - * - * See \ref nvmlDriverModel_t for details on available driver models. - * See \ref nvmlFlagDefault and \ref nvmlFlagForce - * - * @param device The identifier of the target device - * @param driverModel The target driver model - * @param flags Flags that change the default behavior - * - * @return - * - \ref NVML_SUCCESS if the driver model has been set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a driverModel is invalid - * - \ref NVML_ERROR_NOT_SUPPORTED if the platform is not windows or the device does not support this feature - * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * - * @see nvmlDeviceGetDriverModel() - */ -nvmlReturn_t DECLDIR nvmlDeviceSetDriverModel(nvmlDevice_t device, nvmlDriverModel_t driverModel, unsigned int flags); - -/** - * Set clocks that applications will lock to. - * - * Sets the clocks that compute and graphics applications will be running at. - * e.g. CUDA driver requests these clocks during context creation which means this property - * defines clocks at which CUDA applications will be running unless some overspec event - * occurs (e.g. over power, over thermal or external HW brake). - * - * Can be used as a setting to request constant performance. - * - * On Pascal and newer hardware, this will automatically disable automatic boosting of clocks. - * - * On K80 and newer Kepler and Maxwell GPUs, users desiring fixed performance should also call - * \ref nvmlDeviceSetAutoBoostedClocksEnabled to prevent clocks from automatically boosting - * above the clock value being set. - * - * For Kepler &tm; or newer non-GeForce fully supported devices and Maxwell or newer GeForce devices. - * Requires root/admin permissions. - * - * See \ref nvmlDeviceGetSupportedMemoryClocks and \ref nvmlDeviceGetSupportedGraphicsClocks - * for details on how to list available clocks combinations. - * - * After system reboot or driver reload applications clocks go back to their default value. - * See \ref nvmlDeviceResetApplicationsClocks. - * - * @param device The identifier of the target device - * @param memClockMHz Requested memory clock in MHz - * @param graphicsClockMHz Requested graphics clock in MHz - * - * @return - * - \ref NVML_SUCCESS if new settings were successfully set - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a memClockMHz and \a graphicsClockMHz - * is not a valid clock combination - * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation - * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature - * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlDeviceSetApplicationsClocks(nvmlDevice_t device, unsigned int memClockMHz, unsigned int graphicsClockMHz); +nvmlReturn_t DECLDIR nvmlDeviceSetTemperatureThreshold(nvmlDevice_t device, nvmlTemperatureThresholds_t thresholdType, int *temp); /** * Set new power limit of this device. - * + * * For Kepler &tm; or newer fully supported devices. * Requires root/admin permissions. * @@ -4327,8 +6799,8 @@ nvmlReturn_t DECLDIR nvmlDeviceSetApplicationsClocks(nvmlDevice_t device, unsign * * @param device The identifier of the target device * @param limit Power management limit in milliwatts to set - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a limit has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a defaultLimit is out of range @@ -4348,17 +6820,17 @@ nvmlReturn_t DECLDIR nvmlDeviceSetPowerManagementLimit(nvmlDevice_t device, unsi * Modes \ref NVML_GOM_LOW_DP and \ref NVML_GOM_ALL_ON are supported on fully supported GeForce products. * Not supported on Quadro ® and Tesla &tm; C-class products. * Requires root/admin permissions. - * - * Changing GOMs requires a reboot. + * + * Changing GOMs requires a reboot. * The reboot requirement might be removed in the future. * * Compute only GOMs don't support graphics acceleration. Under windows switching to these GOMs when * pending driver model is WDDM is not supported. See \ref nvmlDeviceSetDriverModel. - * + * * @param device The identifier of the target device * @param mode Target GOM - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a mode has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a mode incorrect @@ -4377,7 +6849,7 @@ nvmlReturn_t DECLDIR nvmlDeviceSetGpuOperationMode(nvmlDevice_t device, nvmlGpuO * This method can be used by a root/admin user to give non-root/admin access to certain otherwise-restricted APIs. * The new setting lasts for the lifetime of the NVIDIA driver; it is not persistent. See \a nvmlDeviceGetAPIRestriction * to query the current restriction settings. - * + * * For Kepler &tm; or newer fully supported devices. * Requires root/admin permissions. * @@ -4390,7 +6862,7 @@ nvmlReturn_t DECLDIR nvmlDeviceSetGpuOperationMode(nvmlDevice_t device, nvmlGpuO * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a apiType incorrect * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support changing API restrictions or the device does not support - * the feature that api restrictions are being set for (E.G. Enabling/disabling auto + * the feature that api restrictions are being set for (E.G. Enabling/disabling auto * boosted clocks is not supported by the device) * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible @@ -4400,10 +6872,98 @@ nvmlReturn_t DECLDIR nvmlDeviceSetGpuOperationMode(nvmlDevice_t device, nvmlGpuO */ nvmlReturn_t DECLDIR nvmlDeviceSetAPIRestriction(nvmlDevice_t device, nvmlRestrictedAPI_t apiType, nvmlEnableState_t isRestricted); +/** + * Sets the speed of a specified fan. + * + * WARNING: This function changes the fan control policy to manual. It means that YOU have to monitor + * the temperature and adjust the fan speed accordingly. + * If you set the fan speed too low you can burn your GPU! + * Use nvmlDeviceSetDefaultFanSpeed_v2 to restore default control policy. + * + * For all cuda-capable discrete products with fans that are Maxwell or Newer. + * + * device The identifier of the target device + * fan The index of the fan, starting at zero + * speed The target speed of the fan [0-100] in % of max speed + * + * return + * NVML_SUCCESS if the fan speed has been set + * NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * NVML_ERROR_INVALID_ARGUMENT if the device is not valid, or the speed is outside acceptable ranges, + * or if the fan index doesn't reference an actual fan. + * NVML_ERROR_NOT_SUPPORTED if the device is older than Maxwell. + * NVML_ERROR_UNKNOWN if there was an unexpected error. + */ +nvmlReturn_t DECLDIR nvmlDeviceSetFanSpeed_v2(nvmlDevice_t device, unsigned int fan, unsigned int speed); + +/** + * Set the GPCCLK VF offset value + * @param[in] device The identifier of the target device + * @param[in] offset The GPCCLK VF offset value to set + * + * @return + * - \ref NVML_SUCCESS if \a offset has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a offset is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceSetGpcClkVfOffset(nvmlDevice_t device, int offset); + +/** + * Set the MemClk (Memory Clock) VF offset value. It requires elevated privileges. + * @param[in] device The identifier of the target device + * @param[in] offset The MemClk VF offset value to set + * + * @return + * - \ref NVML_SUCCESS if \a offset has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a offset is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature + * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceSetMemClkVfOffset(nvmlDevice_t device, int offset); + +/** + * Set Conf Computing Unprotected Memory Size. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param device Device Handle + * @param sizeKiB Unprotected Memory size to be set in KiB + * + * @return + * - \ref NVML_SUCCESS if \a sizeKiB successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + */ +nvmlReturn_t DECLDIR nvmlDeviceSetConfComputeUnprotectedMemSize(nvmlDevice_t device, unsigned long long sizeKiB); + +/** + * Set Conf Computing GPUs ready state. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux, Windows TCC. + * + * @param isAcceptingWork GPU accepting new work, NVML_CC_ACCEPTING_CLIENT_REQUESTS_TRUE or + * NVML_CC_ACCEPTING_CLIENT_REQUESTS_FALSE + * + * return + * - \ref NVML_SUCCESS if \a current GPUs ready state is successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a isAcceptingWork is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + */ +nvmlReturn_t DECLDIR nvmlSystemSetConfComputeGpusReadyState(unsigned int isAcceptingWork); + /** * @} */ - + /** @addtogroup nvmlAccountingStats * @{ */ @@ -4416,11 +6976,13 @@ nvmlReturn_t DECLDIR nvmlDeviceSetAPIRestriction(nvmlDevice_t device, nvmlRestri * * @note This setting is not persistent and will default to disabled after driver unloads. * Enable persistence mode to be sure the setting doesn't switch off to disabled. - * + * * @note Enabling accounting mode has no negative impact on the GPU performance. * * @note Disabling accounting clears all accounting pids information. * + * @note On MIG-enabled GPUs, accounting mode would be set to DISABLED and changing it is not supported. + * * See \ref nvmlDeviceGetAccountingMode * See \ref nvmlDeviceGetAccountingStats * See \ref nvmlDeviceClearAccountingPids @@ -4428,7 +6990,7 @@ nvmlReturn_t DECLDIR nvmlDeviceSetAPIRestriction(nvmlDevice_t device, nvmlRestri * @param device The identifier of the target device * @param mode The target accounting mode * - * @return + * @return * - \ref NVML_SUCCESS if the new mode has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a mode are invalid @@ -4450,8 +7012,8 @@ nvmlReturn_t DECLDIR nvmlDeviceSetAccountingMode(nvmlDevice_t device, nvmlEnable * * @param device The identifier of the target device * - * @return - * - \ref NVML_SUCCESS if accounting information has been cleared + * @return + * - \ref NVML_SUCCESS if accounting information has been cleared * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device are invalid * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature @@ -4477,10 +7039,10 @@ nvmlReturn_t DECLDIR nvmlDeviceClearAccountingPids(nvmlDevice_t device); * @param device The identifier of the target device * @param link Specifies the NvLink link to be queried * @param isActive \a nvmlEnableState_t where NVML_FEATURE_ENABLED indicates that - * the link is active and NVML_FEATURE_DISABLED indicates it + * the link is active and NVML_FEATURE_DISABLED indicates it * is inactive * - * @return + * @return * - \ref NVML_SUCCESS if \a isActive has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a link is invalid or \a isActive is NULL @@ -4498,7 +7060,7 @@ nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkState(nvmlDevice_t device, unsigned int * @param link Specifies the NvLink link to be queried * @param version Requested NvLink version * - * @return + * @return * - \ref NVML_SUCCESS if \a version has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a link is invalid or \a version is NULL @@ -4519,7 +7081,7 @@ nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkVersion(nvmlDevice_t device, unsigned in * @param capability Specifies the \a nvmlNvLinkCapability_t to be queried * @param capResult A boolean for the queried capability indicating that feature is available * - * @return + * @return * - \ref NVML_SUCCESS if \a capResult has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a link, or \a capability is invalid or \a capResult is NULL @@ -4527,26 +7089,26 @@ nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkVersion(nvmlDevice_t device, unsigned in * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkCapability(nvmlDevice_t device, unsigned int link, - nvmlNvLinkCapability_t capability, unsigned int *capResult); + nvmlNvLinkCapability_t capability, unsigned int *capResult); /** - * Retrieves the PCI information for the remote node on a NvLink link + * Retrieves the PCI information for the remote node on a NvLink link * Note: pciSubSystemId is not filled in this function and is indeterminate * * For Pascal &tm; or newer fully supported devices. * * @param device The identifier of the target device * @param link Specifies the NvLink link to be queried - * @param pci \a nvmlPciInfo_t of the remote node for the specified link + * @param pci \a nvmlPciInfo_t of the remote node for the specified link * - * @return + * @return * - \ref NVML_SUCCESS if \a pci has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a link is invalid or \a pci is NULL * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ -nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkRemotePciInfo(nvmlDevice_t device, unsigned int link, nvmlPciInfo_t *pci); +nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkRemotePciInfo_v2(nvmlDevice_t device, unsigned int link, nvmlPciInfo_t *pci); /** * Retrieves the specified error counter value @@ -4559,7 +7121,7 @@ nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkRemotePciInfo(nvmlDevice_t device, unsig * @param counter Specifies the NvLink counter to be queried * @param counterValue Returned counter value * - * @return + * @return * - \ref NVML_SUCCESS if \a counter has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a link, or \a counter is invalid or \a counterValue is NULL @@ -4578,7 +7140,7 @@ nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkErrorCounter(nvmlDevice_t device, unsign * @param device The identifier of the target device * @param link Specifies the NvLink link to be queried * - * @return + * @return * - \ref NVML_SUCCESS if the reset is successful * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a link is invalid @@ -4588,6 +7150,8 @@ nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkErrorCounter(nvmlDevice_t device, unsign nvmlReturn_t DECLDIR nvmlDeviceResetNvLinkErrorCounters(nvmlDevice_t device, unsigned int link); /** + * Deprecated: Setting utilization counter control is no longer supported. + * * Set the NVLINK utilization counter control information for the specified counter, 0 or 1. * Please refer to \a nvmlNvLinkUtilizationControl_t for the structure definition. Performs a reset * of the counters if the reset parameter is non-zero. @@ -4600,10 +7164,10 @@ nvmlReturn_t DECLDIR nvmlDeviceResetNvLinkErrorCounters(nvmlDevice_t device, uns * @param control A reference to the \a nvmlNvLinkUtilizationControl_t to set * @param reset Resets the counters on set if non-zero * - * @return + * @return * - \ref NVML_SUCCESS if the control has been set successfully * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a counter, \a link, or \a control is invalid + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a counter, \a link, or \a control is invalid * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ @@ -4611,6 +7175,8 @@ nvmlReturn_t DECLDIR nvmlDeviceSetNvLinkUtilizationControl(nvmlDevice_t device, nvmlNvLinkUtilizationControl_t *control, unsigned int reset); /** + * Deprecated: Getting utilization counter control is no longer supported. + * * Get the NVLINK utilization counter control information for the specified counter, 0 or 1. * Please refer to \a nvmlNvLinkUtilizationControl_t for the structure definition * @@ -4621,10 +7187,10 @@ nvmlReturn_t DECLDIR nvmlDeviceSetNvLinkUtilizationControl(nvmlDevice_t device, * @param link Specifies the NvLink link to be queried * @param control A reference to the \a nvmlNvLinkUtilizationControl_t to place information * - * @return + * @return * - \ref NVML_SUCCESS if the control has been set successfully * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a counter, \a link, or \a control is invalid + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a counter, \a link, or \a control is invalid * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ @@ -4633,6 +7199,8 @@ nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkUtilizationControl(nvmlDevice_t device, /** + * Deprecated: Use \ref nvmlDeviceGetFieldValues with NVML_FI_DEV_NVLINK_THROUGHPUT_* as field values instead. + * * Retrieve the NVLINK utilization counter based on the current control for a specified counter. * In general it is good practice to use \a nvmlDeviceSetNvLinkUtilizationControl * before reading the utilization counters as they have no default state @@ -4645,18 +7213,20 @@ nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkUtilizationControl(nvmlDevice_t device, * @param rxcounter Receive counter return value * @param txcounter Transmit counter return value * - * @return + * @return * - \ref NVML_SUCCESS if \a rxcounter and \a txcounter have been successfully set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a counter, or \a link is invalid or \a rxcounter or \a txcounter are NULL * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ -nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkUtilizationCounter(nvmlDevice_t device, unsigned int link, unsigned int counter, +nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkUtilizationCounter(nvmlDevice_t device, unsigned int link, unsigned int counter, unsigned long long *rxcounter, unsigned long long *txcounter); /** - * Freeze the NVLINK utilization counters + * Deprecated: Freezing NVLINK utilization counters is no longer supported. + * + * Freeze the NVLINK utilization counters * Both the receive and transmit counters are operated on by this function * * For Pascal &tm; or newer fully supported devices. @@ -4667,18 +7237,20 @@ nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkUtilizationCounter(nvmlDevice_t device, * @param freeze NVML_FEATURE_ENABLED = freeze the receive and transmit counters * NVML_FEATURE_DISABLED = unfreeze the receive and transmit counters * - * @return + * @return * - \ref NVML_SUCCESS if counters were successfully frozen or unfrozen * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a link, \a counter, or \a freeze is invalid * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ -nvmlReturn_t DECLDIR nvmlDeviceFreezeNvLinkUtilizationCounter (nvmlDevice_t device, unsigned int link, +nvmlReturn_t DECLDIR nvmlDeviceFreezeNvLinkUtilizationCounter (nvmlDevice_t device, unsigned int link, unsigned int counter, nvmlEnableState_t freeze); /** - * Reset the NVLINK utilization counters + * Deprecated: Resetting NVLINK utilization counters is no longer supported. + * + * Reset the NVLINK utilization counters * Both the receive and transmit counters are operated on by this function * * For Pascal &tm; or newer fully supported devices. @@ -4687,7 +7259,7 @@ nvmlReturn_t DECLDIR nvmlDeviceFreezeNvLinkUtilizationCounter (nvmlDevice_t devi * @param link Specifies the NvLink link to be reset * @param counter Specifies the counter that should be reset (0 or 1) * - * @return + * @return * - \ref NVML_SUCCESS if counters were successfully reset * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a link, or \a counter is invalid @@ -4696,11 +7268,30 @@ nvmlReturn_t DECLDIR nvmlDeviceFreezeNvLinkUtilizationCounter (nvmlDevice_t devi */ nvmlReturn_t DECLDIR nvmlDeviceResetNvLinkUtilizationCounter (nvmlDevice_t device, unsigned int link, unsigned int counter); +/** +* Get the NVLink device type of the remote device connected over the given link. +* +* @param device The device handle of the target GPU +* @param link The NVLink link index on the target GPU +* @param pNvLinkDeviceType Pointer in which the output remote device type is returned +* +* @return +* - \ref NVML_SUCCESS if \a pNvLinkDeviceType has been set +* - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized +* - \ref NVML_ERROR_NOT_SUPPORTED if NVLink is not supported +* - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a link is invalid, or +* \a pNvLinkDeviceType is NULL +* - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is +* otherwise inaccessible +* - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkRemoteDeviceType(nvmlDevice_t device, unsigned int link, nvmlIntNvLinkDeviceType_t *pNvLinkDeviceType); + /** @} */ /***************************************************************************************************/ /** @defgroup nvmlEvents Event Handling Methods - * This chapter describes methods that NVML can perform against each device to register and wait for + * This chapter describes methods that NVML can perform against each device to register and wait for * some event to occur. * @{ */ @@ -4712,13 +7303,13 @@ nvmlReturn_t DECLDIR nvmlDeviceResetNvLinkUtilizationCounter (nvmlDevice_t devic * * For Fermi &tm; or newer fully supported devices. * @param set Reference in which to return the event handle - * - * @return + * + * @return * - \ref NVML_SUCCESS if the event has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a set is NULL * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * + * * @see nvmlEventSetFree */ nvmlReturn_t DECLDIR nvmlEventSetCreate(nvmlEventSet_t *set); @@ -4736,7 +7327,7 @@ nvmlReturn_t DECLDIR nvmlEventSetCreate(nvmlEventSet_t *set); * * This call starts recording of events on specific device. * All events that occurred before this call are not recorded. - * Checking if some event occurred can be done with \ref nvmlEventSetWait + * Checking if some event occurred can be done with \ref nvmlEventSetWait_v2 * * If function reports NVML_ERROR_UNKNOWN, event set is in undefined state and should be freed. * If function reports NVML_ERROR_NOT_SUPPORTED, event set can still be used. None of the requested eventTypes @@ -4745,15 +7336,15 @@ nvmlReturn_t DECLDIR nvmlEventSetCreate(nvmlEventSet_t *set); * @param device The identifier of the target device * @param eventTypes Bitmask of \ref nvmlEventType to record * @param set Set to which add new event types - * - * @return + * + * @return * - \ref NVML_SUCCESS if the event has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a eventTypes is invalid or \a set is NULL * - \ref NVML_ERROR_NOT_SUPPORTED if the platform does not support this feature or some of requested event types * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * + * * @see nvmlEventType * @see nvmlDeviceGetSupportedEventTypes * @see nvmlEventSetWait @@ -4770,14 +7361,14 @@ nvmlReturn_t DECLDIR nvmlDeviceRegisterEvents(nvmlDevice_t device, unsigned long * * @param device The identifier of the target device * @param eventTypes Reference in which to return bitmask of supported events - * - * @return + * + * @return * - \ref NVML_SUCCESS if the eventTypes has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a eventType is NULL * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * + * * @see nvmlEventType * @see nvmlDeviceRegisterEvents */ @@ -4789,43 +7380,51 @@ nvmlReturn_t DECLDIR nvmlDeviceGetSupportedEventTypes(nvmlDevice_t device, unsig * For Fermi &tm; or newer fully supported devices. * * If some events are ready to be delivered at the time of the call, function returns immediately. - * If there are no events ready to be delivered, function sleeps till event arrives + * If there are no events ready to be delivered, function sleeps till event arrives * but not longer than specified timeout. This function in certain conditions can return before * specified timeout passes (e.g. when interrupt arrives) - * - * In case of xid error, the function returns the most recent xid error type seen by the system. If there are multiple - * xid errors generated before nvmlEventSetWait is invoked then the last seen xid error type is returned for all - * xid error events. - * + * + * On Windows, in case of xid error, the function returns the most recent xid error type seen by the system. + * If there are multiple xid errors generated before nvmlEventSetWait is invoked then the last seen xid error + * type is returned for all xid error events. + * + * On Linux, every xid error event would return the associated event data and other information if applicable. + * + * In MIG mode, if device handle is provided, the API reports all the events for the available instances, + * only if the caller has appropriate privileges. In absence of required privileges, only the events which + * affect all the instances (i.e. whole device) are reported. + * + * This API does not currently support per-instance event reporting using MIG device handles. + * * @param set Reference to set of events to wait on * @param data Reference in which to return event data * @param timeoutms Maximum amount of wait time in milliseconds for registered event - * - * @return + * + * @return * - \ref NVML_SUCCESS if the data has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a data is NULL * - \ref NVML_ERROR_TIMEOUT if no event arrived in specified timeout or interrupt arrived * - \ref NVML_ERROR_GPU_IS_LOST if a GPU has fallen off the bus or is otherwise inaccessible * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * + * * @see nvmlEventType * @see nvmlDeviceRegisterEvents */ -nvmlReturn_t DECLDIR nvmlEventSetWait(nvmlEventSet_t set, nvmlEventData_t * data, unsigned int timeoutms); +nvmlReturn_t DECLDIR nvmlEventSetWait_v2(nvmlEventSet_t set, nvmlEventData_t * data, unsigned int timeoutms); /** * Releases events in the set * * For Fermi &tm; or newer fully supported devices. * - * @param set Reference to events to be released - * - * @return + * @param set Reference to events to be released + * + * @return * - \ref NVML_SUCCESS if the event has been successfully released * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_UNKNOWN on any unexpected error - * + * * @see nvmlDeviceRegisterEvents */ nvmlReturn_t DECLDIR nvmlEventSetFree(nvmlEventSet_t set); @@ -4833,7 +7432,7 @@ nvmlReturn_t DECLDIR nvmlEventSetFree(nvmlEventSet_t set); /** @} */ /***************************************************************************************************/ -/** @defgroup nvmlZPI Drain states +/** @defgroup nvmlZPI Drain states * This chapter describes methods that NVML can perform against each device to control their drain state * and recognition by NVML and NVIDIA kernel driver. These methods can be used with out-of-band tools to * power on/off GPUs, enable robust reset scenarios, etc. @@ -4847,14 +7446,14 @@ nvmlReturn_t DECLDIR nvmlEventSetFree(nvmlEventSet_t set); * this call is made. * Must be called as administrator. * For Linux only. - * + * * For Pascal &tm; or newer fully supported devices. * Some Kepler devices supported. * * @param pciInfo The PCI address of the GPU drain state to be modified * @param newState The drain state that should be entered, see \ref nvmlEnableState_t * - * @return + * @return * - \ref NVML_SUCCESS if counters were successfully reset * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a nvmlIndex or \a newState is invalid @@ -4869,14 +7468,14 @@ nvmlReturn_t DECLDIR nvmlDeviceModifyDrainState (nvmlPciInfo_t *pciInfo, nvmlEna * Query the drain state of a GPU. This method is used to check if a GPU is in a currently draining * state. * For Linux only. - * + * * For Pascal &tm; or newer fully supported devices. * Some Kepler devices supported. * * @param pciInfo The PCI address of the GPU drain state to be queried * @param currentState The current drain state for this GPU, see \ref nvmlEnableState_t * - * @return + * @return * - \ref NVML_SUCCESS if counters were successfully reset * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a nvmlIndex or \a currentState is invalid @@ -4890,20 +7489,23 @@ nvmlReturn_t DECLDIR nvmlDeviceQueryDrainState (nvmlPciInfo_t *pciInfo, nvmlEnab * as long as no other processes are attached. If other processes are attached, this call will return * NVML_ERROR_IN_USE and the GPU will be returned to its original "draining" state. Note: the * only situation where a process can still be attached after nvmlDeviceModifyDrainState() is called - * to initiate the draining state is if that process was using, and is still using, a GPU before the + * to initiate the draining state is if that process was using, and is still using, a GPU before the * call was made. Also note, persistence mode counts as an attachment to the GPU thus it must be disabled * prior to this call. * * For long-running NVML processes please note that this will change the enumeration of current GPUs. * For example, if there are four GPUs present and GPU1 is removed, the new enumeration will be 0-2. * Also, device handles after the removed GPU will not be valid and must be re-established. - * Must be run as administrator. + * Must be run as administrator. * For Linux only. * * For Pascal &tm; or newer fully supported devices. * Some Kepler devices supported. * * @param pciInfo The PCI address of the GPU to be removed + * @param gpuState Whether the GPU is to be removed, from the OS + * see \ref nvmlDetachGpuState_t + * @param linkState Requested upstream PCIe link state, see \ref nvmlPcieLinkState_t * * @return * - \ref NVML_SUCCESS if counters were successfully reset @@ -4912,11 +7514,11 @@ nvmlReturn_t DECLDIR nvmlDeviceQueryDrainState (nvmlPciInfo_t *pciInfo, nvmlEnab * - \ref NVML_ERROR_NOT_SUPPORTED if the device doesn't support this feature * - \ref NVML_ERROR_IN_USE if the device is still in use and cannot be removed */ -nvmlReturn_t DECLDIR nvmlDeviceRemoveGpu (nvmlPciInfo_t *pciInfo); +nvmlReturn_t DECLDIR nvmlDeviceRemoveGpu_v2(nvmlPciInfo_t *pciInfo, nvmlDetachGpuState_t gpuState, nvmlPcieLinkState_t linkState); /** * Request the OS and the NVIDIA kernel driver to rediscover a portion of the PCI subsystem looking for GPUs that - * were previously removed. The portion of the PCI tree can be narrowed by specifying a domain, bus, and device. + * were previously removed. The portion of the PCI tree can be narrowed by specifying a domain, bus, and device. * If all are zeroes then the entire PCI tree will be searched. Please note that for long-running NVML processes * the enumeration will change based on how many GPUs are discovered and where they are inserted in bus order. * @@ -4925,14 +7527,14 @@ nvmlReturn_t DECLDIR nvmlDeviceRemoveGpu (nvmlPciInfo_t *pciInfo); * * Must be run as administrator. * For Linux only. - * + * * For Pascal &tm; or newer fully supported devices. * Some Kepler devices supported. * * @param pciInfo The PCI tree to be searched. Only the domain, bus, and device * fields are used in this call. * - * @return + * @return * - \ref NVML_SUCCESS if counters were successfully reset * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a pciInfo is invalid @@ -4970,12 +7572,34 @@ nvmlReturn_t DECLDIR nvmlDeviceDiscoverGpus (nvmlPciInfo_t *pciInfo); */ nvmlReturn_t DECLDIR nvmlDeviceGetFieldValues(nvmlDevice_t device, int valuesCount, nvmlFieldValue_t *values); +/** + * Clear values for a list of fields for a device. This API allows multiple fields to be cleared at once. + * + * @param device The device handle of the GPU to request field values for + * @param valuesCount Number of entries in values that should be cleared + * @param values Array of \a valuesCount structures to hold field values. + * Each value's fieldId must be populated prior to this call + * + * @return + * - \ref NVML_SUCCESS if any values in \a values were cleared. Note that you must + * check the nvmlReturn field of each value for each individual + * status + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a values is NULL + */ +nvmlReturn_t DECLDIR nvmlDeviceClearFieldValues(nvmlDevice_t device, int valuesCount, nvmlFieldValue_t *values); /** @} */ /***************************************************************************************************/ -/** @defgroup nvmlGridQueries Grid Queries - * This chapter describes NVML operations that are associated with NVIDIA GRID products. +/** @defgroup vGPU Enums, Constants and Structs + * @{ + */ +/** @} */ +/***************************************************************************************************/ + +/***************************************************************************************************/ +/** @defgroup nvmlVirtualGpuQueries vGPU APIs + * This chapter describes operations that are associated with NVIDIA vGPU Software products. * @{ */ /***************************************************************************************************/ @@ -4987,8 +7611,8 @@ nvmlReturn_t DECLDIR nvmlDeviceGetFieldValues(nvmlDevice_t device, int valuesCou * * @param device Identifier of the target device * @param pVirtualMode Reference to virtualization mode. One of NVML_GPU_VIRTUALIZATION_? - * - * @return + * + * @return * - \ref NVML_SUCCESS if \a pVirtualMode is fetched * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pVirtualMode is NULL @@ -4997,14 +7621,23 @@ nvmlReturn_t DECLDIR nvmlDeviceGetFieldValues(nvmlDevice_t device, int valuesCou */ nvmlReturn_t DECLDIR nvmlDeviceGetVirtualizationMode(nvmlDevice_t device, nvmlGpuVirtualizationMode_t *pVirtualMode); -/** @} */ - -/***************************************************************************************************/ -/** @defgroup nvmlGridCommands Grid Commands - * This chapter describes NVML operations that are associated with NVIDIA GRID products. - * @{ +/** + * Queries if SR-IOV host operation is supported on a vGPU supported device. + * + * Checks whether SR-IOV host capability is supported by the device and the + * driver, and indicates device is in SR-IOV mode if both of these conditions + * are true. + * + * @param device The identifier of the target device + * @param pHostVgpuMode Reference in which to return the current vGPU mode + * + * @return + * - \ref NVML_SUCCESS if device's vGPU mode has been successfully retrieved + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device handle is 0 or \a pVgpuMode is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if \a device doesn't support this feature. + * - \ref NVML_ERROR_UNKNOWN if any unexpected error occurred */ -/***************************************************************************************************/ +nvmlReturn_t DECLDIR nvmlDeviceGetHostVgpuMode(nvmlDevice_t device, nvmlHostVgpuMode_t *pHostVgpuMode); /** * This method is used to set the virtualization mode corresponding to the GPU. @@ -5014,7 +7647,7 @@ nvmlReturn_t DECLDIR nvmlDeviceGetVirtualizationMode(nvmlDevice_t device, nvmlGp * @param device Identifier of the target device * @param virtualMode virtualization mode. One of NVML_GPU_VIRTUALIZATION_? * - * @return + * @return * - \ref NVML_SUCCESS if \a pVirtualMode is set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a pVirtualMode is NULL @@ -5024,16 +7657,75 @@ nvmlReturn_t DECLDIR nvmlDeviceGetVirtualizationMode(nvmlDevice_t device, nvmlGp */ nvmlReturn_t DECLDIR nvmlDeviceSetVirtualizationMode(nvmlDevice_t device, nvmlGpuVirtualizationMode_t virtualMode); +/** + * Retrieve the vGPU Software licensable features. + * + * Identifies whether the system supports vGPU Software Licensing. If it does, return the list of licensable feature(s) + * and their current license status. + * + * @param device Identifier of the target device + * @param pGridLicensableFeatures Pointer to structure in which vGPU software licensable features are returned + * + * @return + * - \ref NVML_SUCCESS if licensable features are successfully retrieved + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a pGridLicensableFeatures is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGridLicensableFeatures_v4(nvmlDevice_t device, nvmlGridLicensableFeatures_t *pGridLicensableFeatures); + /** @} */ /***************************************************************************************************/ /** @defgroup nvmlVgpu vGPU Management * @{ * - * Set of APIs supporting GRID vGPU + * This chapter describes APIs supporting NVIDIA vGPU. */ /***************************************************************************************************/ +/** + * Retrieve the requested vGPU driver capability. + * + * Refer to the \a nvmlVgpuDriverCapability_t structure for the specific capabilities that can be queried. + * The return value in \a capResult should be treated as a boolean, with a non-zero value indicating that the capability + * is supported. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param capability Specifies the \a nvmlVgpuDriverCapability_t to be queried + * @param capResult A boolean for the queried capability indicating that feature is supported + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a capability is invalid, or \a capResult is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED the API is not supported in current state or \a devices not in vGPU mode + * - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlGetVgpuDriverCapabilities(nvmlVgpuDriverCapability_t capability, unsigned int *capResult); + +/** + * Retrieve the requested vGPU capability for GPU. + * + * Refer to the \a nvmlDeviceVgpuCapability_t structure for the specific capabilities that can be queried. + * The return value in \a capResult reports a non-zero value indicating that the capability + * is supported, and also reports the capability's data based on the queried capability. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param device The identifier of the target device + * @param capability Specifies the \a nvmlDeviceVgpuCapability_t to be queried + * @param capResult Specifies that the queried capability is supported, and also returns capability's data + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a capability is invalid, or \a capResult is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED the API is not supported in current state or \a device not in vGPU mode + * - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlDeviceGetVgpuCapabilities(nvmlDevice_t device, nvmlDeviceVgpuCapability_t capability, unsigned int *capResult); + /** * Retrieve the supported vGPU types on a physical GPU (device). * @@ -5055,8 +7747,7 @@ nvmlReturn_t DECLDIR nvmlDeviceSetVirtualizationMode(nvmlDevice_t device, nvmlGp * - \ref NVML_ERROR_INSUFFICIENT_SIZE \a vgpuTypeIds buffer is too small, array element count is returned in \a vgpuCount * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuCount is NULL or \a device is invalid * - \ref NVML_ERROR_NOT_SUPPORTED if vGPU is not supported by the device - * - \ref NVML_ERROR_VGPU_ECC_NOT_SUPPORTED if ECC is enabled on the device - * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ nvmlReturn_t DECLDIR nvmlDeviceGetSupportedVgpus(nvmlDevice_t device, unsigned int *vgpuCount, nvmlVgpuTypeId_t *vgpuTypeIds); @@ -5073,7 +7764,7 @@ nvmlReturn_t DECLDIR nvmlDeviceGetSupportedVgpus(nvmlDevice_t device, unsigned i * * If the supplied buffer is not large enough to accommodate the vGPU type array, the function returns * NVML_ERROR_INSUFFICIENT_SIZE, with the element count of nvmlVgpuTypeId_t array required in \a vgpuCount. - * To query the number of vGPU types createable for the GPU, call this function with *vgpuCount = 0. + * To query the number of vGPU types that can be created for the GPU, call this function with *vgpuCount = 0. * The code will return NVML_ERROR_INSUFFICIENT_SIZE, or NVML_SUCCESS if no vGPU types are creatable. * * @param device The identifier of the target device @@ -5085,7 +7776,6 @@ nvmlReturn_t DECLDIR nvmlDeviceGetSupportedVgpus(nvmlDevice_t device, unsigned i * - \ref NVML_ERROR_INSUFFICIENT_SIZE \a vgpuTypeIds buffer is too small, array element count is returned in \a vgpuCount * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuCount is NULL * - \ref NVML_ERROR_NOT_SUPPORTED if vGPU is not supported by the device - * - \ref NVML_ERROR_VGPU_ECC_NOT_SUPPORTED if ECC is enabled on the device * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ nvmlReturn_t DECLDIR nvmlDeviceGetCreatableVgpus(nvmlDevice_t device, unsigned int *vgpuCount, nvmlVgpuTypeId_t *vgpuTypeIds); @@ -5129,6 +7819,24 @@ nvmlReturn_t DECLDIR nvmlVgpuTypeGetClass(nvmlVgpuTypeId_t vgpuTypeId, char *vgp */ nvmlReturn_t DECLDIR nvmlVgpuTypeGetName(nvmlVgpuTypeId_t vgpuTypeId, char *vgpuTypeName, unsigned int *size); +/** + * Retrieve the GPU Instance Profile ID for the given vGPU type ID. + * The API will return a valid GPU Instance Profile ID for the MIG capable vGPU types, else INVALID_GPU_INSTANCE_PROFILE_ID is + * returned. + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuTypeId Handle to vGPU type + * @param gpuInstanceProfileId GPU Instance Profile ID + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_NOT_SUPPORTED if \a device is not in vGPU Host virtualization mode + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuTypeId is invalid, or \a gpuInstanceProfileId is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetGpuInstanceProfileId(nvmlVgpuTypeId_t vgpuTypeId, unsigned int *gpuInstanceProfileId); + /** * Retrieve the device ID of a vGPU type. * @@ -5233,7 +7941,7 @@ nvmlReturn_t DECLDIR nvmlVgpuTypeGetLicense(nvmlVgpuTypeId_t vgpuTypeId, char *v * - \ref NVML_SUCCESS successful completion * - \ref NVML_ERROR_NOT_SUPPORTED if frame rate limiter is turned off for the vGPU type * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, or \a frameRateLimit is NULL + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuTypeId is invalid, or \a frameRateLimit is NULL * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ nvmlReturn_t DECLDIR nvmlVgpuTypeGetFrameRateLimit(nvmlVgpuTypeId_t vgpuTypeId, unsigned int *frameRateLimit); @@ -5256,11 +7964,26 @@ nvmlReturn_t DECLDIR nvmlVgpuTypeGetFrameRateLimit(nvmlVgpuTypeId_t vgpuTypeId, */ nvmlReturn_t DECLDIR nvmlVgpuTypeGetMaxInstances(nvmlDevice_t device, nvmlVgpuTypeId_t vgpuTypeId, unsigned int *vgpuInstanceCount); +/** + * Retrieve the maximum number of vGPU instances supported per VM for given vGPU type + * + * For Kepler &tm; or newer fully supported devices. + * + * @param vgpuTypeId Handle to vGPU type + * @param vgpuInstanceCountPerVm Pointer to get the max number of vGPU instances supported per VM for given \a vgpuTypeId + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuTypeId is invalid, or \a vgpuInstanceCountPerVm is NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetMaxInstancesPerVm(nvmlVgpuTypeId_t vgpuTypeId, unsigned int *vgpuInstanceCountPerVm); + /** * Retrieve the active vGPU instances on a device. * * An array of active vGPU instances is returned in the caller-supplied buffer pointed at by \a vgpuInstances. The - * array elememt count is passed in \a vgpuCount, and \a vgpuCount is used to return the number of vGPU instances + * array element count is passed in \a vgpuCount, and \a vgpuCount is used to return the number of vGPU instances * written to the buffer. * * If the supplied buffer is not large enough to accommodate the vGPU instance array, the function returns @@ -5303,7 +8026,8 @@ nvmlReturn_t DECLDIR nvmlDeviceGetActiveVgpus(nvmlDevice_t device, unsigned int * @return * - \ref NVML_SUCCESS successful completion * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is invalid, or \a vmId or \a vmIdType are NULL + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vmId or \a vmIdType is NULL, or \a vgpuInstance is 0 + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a size is too small * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ @@ -5325,7 +8049,8 @@ nvmlReturn_t DECLDIR nvmlVgpuInstanceGetVmID(nvmlVgpuInstance_t vgpuInstance, ch * @return * - \ref NVML_SUCCESS successful completion * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is invalid, or \a uuid is NULL + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a uuid is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a size is too small * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ @@ -5339,7 +8064,7 @@ nvmlReturn_t DECLDIR nvmlVgpuInstanceGetUUID(nvmlVgpuInstance_t vgpuInstance, ch * See \ref nvmlConstants::NVML_SYSTEM_DRIVER_VERSION_BUFFER_SIZE. * * nvmlVgpuInstanceGetVmDriverVersion() may be called at any time for a vGPU instance. The guest VM driver version is - * returned as "Unknown" if no NVIDIA driver is installed in the VM, or the VM has not yet booted to the point where the + * returned as "Not Available" if no NVIDIA driver is installed in the VM, or the VM has not yet booted to the point where the * NVIDIA driver is loaded and initialized. * * For Kepler &tm; or newer fully supported devices. @@ -5351,7 +8076,8 @@ nvmlReturn_t DECLDIR nvmlVgpuInstanceGetUUID(nvmlVgpuInstance_t vgpuInstance, ch * @return * - \ref NVML_SUCCESS if \a version has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is invalid + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0 + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ @@ -5370,12 +8096,15 @@ nvmlReturn_t DECLDIR nvmlVgpuInstanceGetVmDriverVersion(nvmlVgpuInstance_t vgpuI * @return * - \ref NVML_SUCCESS successful completion * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is invalid, or \a fbUsage is NULL + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a fbUsage is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ nvmlReturn_t DECLDIR nvmlVgpuInstanceGetFbUsage(nvmlVgpuInstance_t vgpuInstance, unsigned long long *fbUsage); /** + * @deprecated Use \ref nvmlVgpuInstanceGetLicenseInfo_v2. + * * Retrieve the current licensing state of the vGPU instance. * * If the vGPU is currently licensed, \a licensed is set to 1, otherwise it is set to 0. @@ -5388,7 +8117,8 @@ nvmlReturn_t DECLDIR nvmlVgpuInstanceGetFbUsage(nvmlVgpuInstance_t vgpuInstance, * @return * - \ref NVML_SUCCESS if \a licensed has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is invalid, or \a licensed is NULL + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a licensed is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ nvmlReturn_t DECLDIR nvmlVgpuInstanceGetLicenseStatus(nvmlVgpuInstance_t vgpuInstance, unsigned int *licensed); @@ -5406,7 +8136,8 @@ nvmlReturn_t DECLDIR nvmlVgpuInstanceGetLicenseStatus(nvmlVgpuInstance_t vgpuIns * @return * - \ref NVML_SUCCESS if \a vgpuTypeId has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is invalid, or \a vgpuTypeId is NULL + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a vgpuTypeId is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ nvmlReturn_t DECLDIR nvmlVgpuInstanceGetType(nvmlVgpuInstance_t vgpuInstance, nvmlVgpuTypeId_t *vgpuTypeId); @@ -5425,13 +8156,30 @@ nvmlReturn_t DECLDIR nvmlVgpuInstanceGetType(nvmlVgpuInstance_t vgpuInstance, nv * - \ref NVML_SUCCESS if \a frameRateLimit has been set * - \ref NVML_ERROR_NOT_SUPPORTED if frame rate limiter is turned off for the vGPU type * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is invalid, or \a frameRateLimit is NULL + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a frameRateLimit is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ nvmlReturn_t DECLDIR nvmlVgpuInstanceGetFrameRateLimit(nvmlVgpuInstance_t vgpuInstance, unsigned int *frameRateLimit); /** - * Retrieve the encoder Capacity of a vGPU instance, in macroblocks per second. + * Retrieve the current ECC mode of vGPU instance. + * + * @param vgpuInstance The identifier of the target vGPU instance + * @param eccMode Reference in which to return the current ECC mode + * + * @return + * - \ref NVML_SUCCESS if the vgpuInstance's ECC mode has been successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a mode is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_NOT_SUPPORTED if the vGPU doesn't support this feature + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetEccMode(nvmlVgpuInstance_t vgpuInstance, nvmlEnableState_t *eccMode); + +/** + * Retrieve the encoder capacity of a vGPU instance, as a percentage of maximum encoder capacity with valid values in the range 0-100. * * For Maxwell &tm; or newer fully supported devices. * @@ -5441,13 +8189,14 @@ nvmlReturn_t DECLDIR nvmlVgpuInstanceGetFrameRateLimit(nvmlVgpuInstance_t vgpuIn * @return * - \ref NVML_SUCCESS if \a encoderCapacity has been retrieved * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is invalid, or \a encoderQueryType is invalid + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a encoderQueryType is invalid + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ nvmlReturn_t DECLDIR nvmlVgpuInstanceGetEncoderCapacity(nvmlVgpuInstance_t vgpuInstance, unsigned int *encoderCapacity); /** - * Set the encoder Capacity of a vGPU instance, in macroblocks per second. + * Set the encoder capacity of a vGPU instance, as a percentage of maximum encoder capacity with valid values in the range 0-100. * * For Maxwell &tm; or newer fully supported devices. * @@ -5457,11 +8206,525 @@ nvmlReturn_t DECLDIR nvmlVgpuInstanceGetEncoderCapacity(nvmlVgpuInstance_t vgpuI * @return * - \ref NVML_SUCCESS if \a encoderCapacity has been set * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is invalid + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a encoderCapacity is out of range of 0-100. + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ nvmlReturn_t DECLDIR nvmlVgpuInstanceSetEncoderCapacity(nvmlVgpuInstance_t vgpuInstance, unsigned int encoderCapacity); +/** + * Retrieves the current encoder statistics of a vGPU Instance + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param sessionCount Reference to an unsigned int for count of active encoder sessions + * @param averageFps Reference to an unsigned int for trailing average FPS of all active sessions + * @param averageLatency Reference to an unsigned int for encode latency in microseconds + * + * @return + * - \ref NVML_SUCCESS if \a sessionCount, \a averageFps and \a averageLatency is fetched + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a sessionCount , or \a averageFps or \a averageLatency is NULL + * or \a vgpuInstance is 0. + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetEncoderStats(nvmlVgpuInstance_t vgpuInstance, unsigned int *sessionCount, + unsigned int *averageFps, unsigned int *averageLatency); + +/** + * Retrieves information about all active encoder sessions on a vGPU Instance. + * + * An array of active encoder sessions is returned in the caller-supplied buffer pointed at by \a sessionInfo. The + * array element count is passed in \a sessionCount, and \a sessionCount is used to return the number of sessions + * written to the buffer. + * + * If the supplied buffer is not large enough to accommodate the active session array, the function returns + * NVML_ERROR_INSUFFICIENT_SIZE, with the element count of nvmlEncoderSessionInfo_t array required in \a sessionCount. + * To query the number of active encoder sessions, call this function with *sessionCount = 0. The code will return + * NVML_SUCCESS with number of active encoder sessions updated in *sessionCount. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param sessionCount Reference to caller supplied array size, and returns + * the number of sessions. + * @param sessionInfo Reference to caller supplied array in which the list + * of session information us returned. + * + * @return + * - \ref NVML_SUCCESS if \a sessionInfo is fetched + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a sessionCount is too small, array element count is + returned in \a sessionCount + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a sessionCount is NULL, or \a vgpuInstance is 0. + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetEncoderSessions(nvmlVgpuInstance_t vgpuInstance, unsigned int *sessionCount, nvmlEncoderSessionInfo_t *sessionInfo); + +/** +* Retrieves the active frame buffer capture sessions statistics of a vGPU Instance +* +* For Maxwell &tm; or newer fully supported devices. +* +* @param vgpuInstance Identifier of the target vGPU instance +* @param fbcStats Reference to nvmlFBCStats_t structure containing NvFBC stats +* +* @return +* - \ref NVML_SUCCESS if \a fbcStats is fetched +* - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized +* - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a fbcStats is NULL +* - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system +* - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetFBCStats(nvmlVgpuInstance_t vgpuInstance, nvmlFBCStats_t *fbcStats); + +/** +* Retrieves information about active frame buffer capture sessions on a vGPU Instance. +* +* An array of active FBC sessions is returned in the caller-supplied buffer pointed at by \a sessionInfo. The +* array element count is passed in \a sessionCount, and \a sessionCount is used to return the number of sessions +* written to the buffer. +* +* If the supplied buffer is not large enough to accommodate the active session array, the function returns +* NVML_ERROR_INSUFFICIENT_SIZE, with the element count of nvmlFBCSessionInfo_t array required in \a sessionCount. +* To query the number of active FBC sessions, call this function with *sessionCount = 0. The code will return +* NVML_SUCCESS with number of active FBC sessions updated in *sessionCount. +* +* For Maxwell &tm; or newer fully supported devices. +* +* @note hResolution, vResolution, averageFPS and averageLatency data for a FBC session returned in \a sessionInfo may +* be zero if there are no new frames captured since the session started. +* +* @param vgpuInstance Identifier of the target vGPU instance +* @param sessionCount Reference to caller supplied array size, and returns the number of sessions. +* @param sessionInfo Reference in which to return the session information +* +* @return +* - \ref NVML_SUCCESS if \a sessionInfo is fetched +* - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized +* - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a sessionCount is NULL. +* - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system +* - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a sessionCount is too small, array element count is returned in \a sessionCount +* - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetFBCSessions(nvmlVgpuInstance_t vgpuInstance, unsigned int *sessionCount, nvmlFBCSessionInfo_t *sessionInfo); + +/** +* Retrieve the GPU Instance ID for the given vGPU Instance. +* The API will return a valid GPU Instance ID for MIG backed vGPU Instance, else INVALID_GPU_INSTANCE_ID is returned. +* +* For Kepler &tm; or newer fully supported devices. +* +* @param vgpuInstance Identifier of the target vGPU instance +* @param gpuInstanceId GPU Instance ID +* +* @return +* - \ref NVML_SUCCESS successful completion +* - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized +* - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a gpuInstanceId is NULL. +* - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system +* - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetGpuInstanceId(nvmlVgpuInstance_t vgpuInstance, unsigned int *gpuInstanceId); + +/** +* Retrieves the PCI Id of the given vGPU Instance i.e. the PCI Id of the GPU as seen inside the VM. +* +* The vGPU PCI id is returned as "00000000:00:00.0" if NVIDIA driver is not installed on the vGPU instance. +* +* @param vgpuInstance Identifier of the target vGPU instance +* @param vgpuPciId Caller-supplied buffer to return vGPU PCI Id string +* @param length Size of the vgpuPciId buffer +* +* @return +* - \ref NVML_SUCCESS if vGPU PCI Id is sucessfully retrieved +* - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized +* - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a vgpuPciId is NULL +* - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system +* - \ref NVML_ERROR_DRIVER_NOT_LOADED if NVIDIA driver is not running on the vGPU instance +* - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a length is too small, \a length is set to required length +* - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetGpuPciId(nvmlVgpuInstance_t vgpuInstance, char *vgpuPciId, unsigned int *length); + +/** +* Retrieve the requested capability for a given vGPU type. Refer to the \a nvmlVgpuCapability_t structure +* for the specific capabilities that can be queried. The return value in \a capResult should be treated as +* a boolean, with a non-zero value indicating that the capability is supported. +* +* For Maxwell &tm; or newer fully supported devices. +* +* @param vgpuTypeId Handle to vGPU type +* @param capability Specifies the \a nvmlVgpuCapability_t to be queried +* @param capResult A boolean for the queried capability indicating that feature is supported +* +* @return +* - \ref NVML_SUCCESS successful completion +* - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized +* - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuTypeId is invalid, or \a capability is invalid, or \a capResult is NULL +* - \ref NVML_ERROR_UNKNOWN on any unexpected error +*/ +nvmlReturn_t DECLDIR nvmlVgpuTypeGetCapabilities(nvmlVgpuTypeId_t vgpuTypeId, nvmlVgpuCapability_t capability, unsigned int *capResult); + +/** + * Retrieve the MDEV UUID of a vGPU instance. + * + * The MDEV UUID is a globally unique identifier of the mdev device assigned to the VM, and is returned as a 5-part hexadecimal string, + * not exceeding 80 characters in length (including the NULL terminator). + * MDEV UUID is displayed only on KVM platform. + * See \ref nvmlConstants::NVML_DEVICE_UUID_BUFFER_SIZE. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param mdevUuid Pointer to caller-supplied buffer to hold MDEV UUID + * @param size Size of buffer in bytes + * + * @return + * - \ref NVML_SUCCESS successful completion + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_NOT_SUPPORTED on any hypervisor other than KVM + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a mdevUuid is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a size is too small + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetMdevUUID(nvmlVgpuInstance_t vgpuInstance, char *mdevUuid, unsigned int size); + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvml vGPU Migration + * This chapter describes operations that are associated with vGPU Migration. + * @{ + */ +/***************************************************************************************************/ + +/** + * Structure representing range of vGPU versions. + */ +typedef struct nvmlVgpuVersion_st +{ + unsigned int minVersion; //!< Minimum vGPU version. + unsigned int maxVersion; //!< Maximum vGPU version. +} nvmlVgpuVersion_t; + +/** + * vGPU metadata structure. + */ +typedef struct nvmlVgpuMetadata_st +{ + unsigned int version; //!< Current version of the structure + unsigned int revision; //!< Current revision of the structure + nvmlVgpuGuestInfoState_t guestInfoState; //!< Current state of Guest-dependent fields + char guestDriverVersion[NVML_SYSTEM_DRIVER_VERSION_BUFFER_SIZE]; //!< Version of driver installed in guest + char hostDriverVersion[NVML_SYSTEM_DRIVER_VERSION_BUFFER_SIZE]; //!< Version of driver installed in host + unsigned int reserved[6]; //!< Reserved for internal use + unsigned int vgpuVirtualizationCaps; //!< vGPU virtualization capabilities bitfield + unsigned int guestVgpuVersion; //!< vGPU version of guest driver + unsigned int opaqueDataSize; //!< Size of opaque data field in bytes + char opaqueData[4]; //!< Opaque data +} nvmlVgpuMetadata_t; + +/** + * Physical GPU metadata structure + */ +typedef struct nvmlVgpuPgpuMetadata_st +{ + unsigned int version; //!< Current version of the structure + unsigned int revision; //!< Current revision of the structure + char hostDriverVersion[NVML_SYSTEM_DRIVER_VERSION_BUFFER_SIZE]; //!< Host driver version + unsigned int pgpuVirtualizationCaps; //!< Pgpu virtualization capabilities bitfield + unsigned int reserved[5]; //!< Reserved for internal use + nvmlVgpuVersion_t hostSupportedVgpuRange; //!< vGPU version range supported by host driver + unsigned int opaqueDataSize; //!< Size of opaque data field in bytes + char opaqueData[4]; //!< Opaque data +} nvmlVgpuPgpuMetadata_t; + +/** + * vGPU VM compatibility codes + */ +typedef enum nvmlVgpuVmCompatibility_enum +{ + NVML_VGPU_VM_COMPATIBILITY_NONE = 0x0, //!< vGPU is not runnable + NVML_VGPU_VM_COMPATIBILITY_COLD = 0x1, //!< vGPU is runnable from a cold / powered-off state (ACPI S5) + NVML_VGPU_VM_COMPATIBILITY_HIBERNATE = 0x2, //!< vGPU is runnable from a hibernated state (ACPI S4) + NVML_VGPU_VM_COMPATIBILITY_SLEEP = 0x4, //!< vGPU is runnable from a sleeped state (ACPI S3) + NVML_VGPU_VM_COMPATIBILITY_LIVE = 0x8 //!< vGPU is runnable from a live/paused (ACPI S0) +} nvmlVgpuVmCompatibility_t; + +/** + * vGPU-pGPU compatibility limit codes + */ +typedef enum nvmlVgpuPgpuCompatibilityLimitCode_enum +{ + NVML_VGPU_COMPATIBILITY_LIMIT_NONE = 0x0, //!< Compatibility is not limited. + NVML_VGPU_COMPATIBILITY_LIMIT_HOST_DRIVER = 0x1, //!< ompatibility is limited by host driver version. + NVML_VGPU_COMPATIBILITY_LIMIT_GUEST_DRIVER = 0x2, //!< Compatibility is limited by guest driver version. + NVML_VGPU_COMPATIBILITY_LIMIT_GPU = 0x4, //!< Compatibility is limited by GPU hardware. + NVML_VGPU_COMPATIBILITY_LIMIT_OTHER = 0x80000000 //!< Compatibility is limited by an undefined factor. +} nvmlVgpuPgpuCompatibilityLimitCode_t; + +/** + * vGPU-pGPU compatibility structure + */ +typedef struct nvmlVgpuPgpuCompatibility_st +{ + nvmlVgpuVmCompatibility_t vgpuVmCompatibility; //!< Compatibility of vGPU VM. See \ref nvmlVgpuVmCompatibility_t + nvmlVgpuPgpuCompatibilityLimitCode_t compatibilityLimitCode; //!< Limiting factor for vGPU-pGPU compatibility. See \ref nvmlVgpuPgpuCompatibilityLimitCode_t +} nvmlVgpuPgpuCompatibility_t; + +/** + * Returns vGPU metadata structure for a running vGPU. The structure contains information about the vGPU and its associated VM + * such as the currently installed NVIDIA guest driver version, together with host driver version and an opaque data section + * containing internal state. + * + * nvmlVgpuInstanceGetMetadata() may be called at any time for a vGPU instance. Some fields in the returned structure are + * dependent on information obtained from the guest VM, which may not yet have reached a state where that information + * is available. The current state of these dependent fields is reflected in the info structure's \ref nvmlVgpuGuestInfoState_t field. + * + * The VMM may choose to read and save the vGPU's VM info as persistent metadata associated with the VM, and provide + * it to Virtual GPU Manager when creating a vGPU for subsequent instances of the VM. + * + * The caller passes in a buffer via \a vgpuMetadata, with the size of the buffer in \a bufferSize. If the vGPU Metadata structure + * is too large to fit in the supplied buffer, the function returns NVML_ERROR_INSUFFICIENT_SIZE with the size needed + * in \a bufferSize. + * + * @param vgpuInstance vGPU instance handle + * @param vgpuMetadata Pointer to caller-supplied buffer into which vGPU metadata is written + * @param bufferSize Size of vgpuMetadata buffer + * + * @return + * - \ref NVML_SUCCESS vGPU metadata structure was successfully returned + * - \ref NVML_ERROR_INSUFFICIENT_SIZE vgpuMetadata buffer is too small, required size is returned in \a bufferSize + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a bufferSize is NULL or \a vgpuInstance is 0; if \a vgpuMetadata is NULL and the value of \a bufferSize is not 0. + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetMetadata(nvmlVgpuInstance_t vgpuInstance, nvmlVgpuMetadata_t *vgpuMetadata, unsigned int *bufferSize); + +/** + * Returns a vGPU metadata structure for the physical GPU indicated by \a device. The structure contains information about + * the GPU and the currently installed NVIDIA host driver version that's controlling it, together with an opaque data section + * containing internal state. + * + * The caller passes in a buffer via \a pgpuMetadata, with the size of the buffer in \a bufferSize. If the \a pgpuMetadata + * structure is too large to fit in the supplied buffer, the function returns NVML_ERROR_INSUFFICIENT_SIZE with the size needed + * in \a bufferSize. + * + * @param device The identifier of the target device + * @param pgpuMetadata Pointer to caller-supplied buffer into which \a pgpuMetadata is written + * @param bufferSize Pointer to size of \a pgpuMetadata buffer + * + * @return + * - \ref NVML_SUCCESS GPU metadata structure was successfully returned + * - \ref NVML_ERROR_INSUFFICIENT_SIZE pgpuMetadata buffer is too small, required size is returned in \a bufferSize + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a bufferSize is NULL or \a device is invalid; if \a pgpuMetadata is NULL and the value of \a bufferSize is not 0. + * - \ref NVML_ERROR_NOT_SUPPORTED vGPU is not supported by the system + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVgpuMetadata(nvmlDevice_t device, nvmlVgpuPgpuMetadata_t *pgpuMetadata, unsigned int *bufferSize); + +/** + * Takes a vGPU instance metadata structure read from \ref nvmlVgpuInstanceGetMetadata(), and a vGPU metadata structure for a + * physical GPU read from \ref nvmlDeviceGetVgpuMetadata(), and returns compatibility information of the vGPU instance and the + * physical GPU. + * + * The caller passes in a buffer via \a compatibilityInfo, into which a compatibility information structure is written. The + * structure defines the states in which the vGPU / VM may be booted on the physical GPU. If the vGPU / VM compatibility + * with the physical GPU is limited, a limit code indicates the factor limiting compatability. + * (see \ref nvmlVgpuPgpuCompatibilityLimitCode_t for details). + * + * Note: vGPU compatibility does not take into account dynamic capacity conditions that may limit a system's ability to + * boot a given vGPU or associated VM. + * + * @param vgpuMetadata Pointer to caller-supplied vGPU metadata structure + * @param pgpuMetadata Pointer to caller-supplied GPU metadata structure + * @param compatibilityInfo Pointer to caller-supplied buffer to hold compatibility info + * + * @return + * - \ref NVML_SUCCESS vGPU metadata structure was successfully returned + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuMetadata or \a pgpuMetadata or \a bufferSize are NULL + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlGetVgpuCompatibility(nvmlVgpuMetadata_t *vgpuMetadata, nvmlVgpuPgpuMetadata_t *pgpuMetadata, nvmlVgpuPgpuCompatibility_t *compatibilityInfo); + +/** + * Returns the properties of the physical GPU indicated by the device in an ascii-encoded string format. + * + * The caller passes in a buffer via \a pgpuMetadata, with the size of the buffer in \a bufferSize. If the + * string is too large to fit in the supplied buffer, the function returns NVML_ERROR_INSUFFICIENT_SIZE with the size needed + * in \a bufferSize. + * + * @param device The identifier of the target device + * @param pgpuMetadata Pointer to caller-supplied buffer into which \a pgpuMetadata is written + * @param bufferSize Pointer to size of \a pgpuMetadata buffer + * + * @return + * - \ref NVML_SUCCESS GPU metadata structure was successfully returned + * - \ref NVML_ERROR_INSUFFICIENT_SIZE \a pgpuMetadata buffer is too small, required size is returned in \a bufferSize + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a bufferSize is NULL or \a device is invalid; if \a pgpuMetadata is NULL and the value of \a bufferSize is not 0. + * - \ref NVML_ERROR_NOT_SUPPORTED if vGPU is not supported by the system + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetPgpuMetadataString(nvmlDevice_t device, char *pgpuMetadata, unsigned int *bufferSize); + +/** + * Returns the vGPU Software scheduler logs. + * \a pSchedulerLog points to a caller-allocated structure to contain the logs. The number of elements returned will + * never exceed \a NVML_SCHEDULER_SW_MAX_LOG_ENTRIES. + * + * To get the entire logs, call the function atleast 5 times a second. + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target \a device + * @param pSchedulerLog Reference in which \a pSchedulerLog is written + * + * @return + * - \ref NVML_SUCCESS vGPU scheduler logs were successfully obtained + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a pSchedulerLog is NULL or \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED The API is not supported in current state or \a device not in vGPU host mode + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVgpuSchedulerLog(nvmlDevice_t device, nvmlVgpuSchedulerLog_t *pSchedulerLog); + +/** + * Returns the vGPU scheduler state. + * The information returned in \a nvmlVgpuSchedulerGetState_t is not relevant if the BEST EFFORT policy is set. + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target \a device + * @param pSchedulerState Reference in which \a pSchedulerState is returned + * + * @return + * - \ref NVML_SUCCESS vGPU scheduler state is successfully obtained + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a pSchedulerState is NULL or \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED The API is not supported in current state or \a device not in vGPU host mode + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVgpuSchedulerState(nvmlDevice_t device, nvmlVgpuSchedulerGetState_t *pSchedulerState); + +/** + * Returns the vGPU scheduler capabilities. + * The list of supported vGPU schedulers returned in \a nvmlVgpuSchedulerCapabilities_t is from + * the NVML_VGPU_SCHEDULER_POLICY_*. This list enumerates the supported scheduler policies + * if the engine is Graphics type. + * The other values in \a nvmlVgpuSchedulerCapabilities_t are also applicable if the engine is + * Graphics type. For other engine types, it is BEST EFFORT policy. + * If ARR is supported and enabled, scheduling frequency and averaging factor are applicable + * else timeSlice is applicable. + * + * For Pascal &tm; or newer fully supported devices. + * + * @param device The identifier of the target \a device + * @param pCapabilities Reference in which \a pCapabilities is written + * + * @return + * - \ref NVML_SUCCESS vGPU scheduler capabilities were successfully obtained + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a pCapabilities is NULL or \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED The API is not supported in current state or \a device not in vGPU host mode + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetVgpuSchedulerCapabilities(nvmlDevice_t device, nvmlVgpuSchedulerCapabilities_t *pCapabilities); + +/** + * Sets the vGPU scheduler state. + * + * For Pascal &tm; or newer fully supported devices. + * + * The scheduler state change won't persist across module load/unload. + * Scheduler state and params will be allowed to set only when no VM is running. + * In \a nvmlVgpuSchedulerSetState_t, IFF enableARRMode is enabled then + * provide avgFactorForARR and frequency as input. If enableARRMode is disabled + * then provide timeslice as input. + * + * @param device The identifier of the target \a device + * @param pSchedulerState vGPU \a pSchedulerState to set + * + * @return + * - \ref NVML_SUCCESS vGPU scheduler state has been successfully set + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a pSchedulerState is NULL or \a device is invalid + * - \ref NVML_ERROR_RESET_REQUIRED if setting \a pSchedulerState failed with fatal error, + * reboot is required to overcome from this error. + * - \ref NVML_ERROR_NOT_SUPPORTED The API is not supported in current state or \a device not in vGPU host mode + * or if any vGPU instance currently exists on the \a device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceSetVgpuSchedulerState(nvmlDevice_t device, nvmlVgpuSchedulerSetState_t *pSchedulerState); + +/* + * Virtual GPU (vGPU) version + * + * The NVIDIA vGPU Manager and the guest drivers are tagged with a range of supported vGPU versions. This determines the range of NVIDIA guest driver versions that + * are compatible for vGPU feature support with a given NVIDIA vGPU Manager. For vGPU feature support, the range of supported versions for the NVIDIA vGPU Manager + * and the guest driver must overlap. Otherwise, the guest driver fails to load in the VM. + * + * When the NVIDIA guest driver loads, either when the VM is booted or when the driver is installed or upgraded, a negotiation occurs between the guest driver + * and the NVIDIA vGPU Manager to select the highest mutually compatible vGPU version. The negotiated vGPU version stays the same across VM migration. + */ + +/** + * Query the ranges of supported vGPU versions. + * + * This function gets the linear range of supported vGPU versions that is preset for the NVIDIA vGPU Manager and the range set by an administrator. + * If the preset range has not been overridden by \ref nvmlSetVgpuVersion, both ranges are the same. + * + * The caller passes pointers to the following \ref nvmlVgpuVersion_t structures, into which the NVIDIA vGPU Manager writes the ranges: + * 1. \a supported structure that represents the preset range of vGPU versions supported by the NVIDIA vGPU Manager. + * 2. \a current structure that represents the range of supported vGPU versions set by an administrator. By default, this range is the same as the preset range. + * + * @param supported Pointer to the structure in which the preset range of vGPU versions supported by the NVIDIA vGPU Manager is written + * @param current Pointer to the structure in which the range of supported vGPU versions set by an administrator is written + * + * @return + * - \ref NVML_SUCCESS The vGPU version range structures were successfully obtained. + * - \ref NVML_ERROR_NOT_SUPPORTED The API is not supported. + * - \ref NVML_ERROR_INVALID_ARGUMENT The \a supported parameter or the \a current parameter is NULL. + * - \ref NVML_ERROR_UNKNOWN An error occurred while the data was being fetched. + */ +nvmlReturn_t DECLDIR nvmlGetVgpuVersion(nvmlVgpuVersion_t *supported, nvmlVgpuVersion_t *current); + +/** + * Override the preset range of vGPU versions supported by the NVIDIA vGPU Manager with a range set by an administrator. + * + * This function configures the NVIDIA vGPU Manager with a range of supported vGPU versions set by an administrator. This range must be a subset of the + * preset range that the NVIDIA vGPU Manager supports. The custom range set by an administrator takes precedence over the preset range and is advertised to + * the guest VM for negotiating the vGPU version. See \ref nvmlGetVgpuVersion for details of how to query the preset range of versions supported. + * + * This function takes a pointer to vGPU version range structure \ref nvmlVgpuVersion_t as input to override the preset vGPU version range that the NVIDIA vGPU Manager supports. + * + * After host system reboot or driver reload, the range of supported versions reverts to the range that is preset for the NVIDIA vGPU Manager. + * + * @note 1. The range set by the administrator must be a subset of the preset range that the NVIDIA vGPU Manager supports. Otherwise, an error is returned. + * 2. If the range of supported guest driver versions does not overlap the range set by the administrator, the guest driver fails to load. + * 3. If the range of supported guest driver versions overlaps the range set by the administrator, the guest driver will load with a negotiated + * vGPU version that is the maximum value in the overlapping range. + * 4. No VMs must be running on the host when this function is called. If a VM is running on the host, the call to this function fails. + * + * @param vgpuVersion Pointer to a caller-supplied range of supported vGPU versions. + * + * @return + * - \ref NVML_SUCCESS The preset range of supported vGPU versions was successfully overridden. + * - \ref NVML_ERROR_NOT_SUPPORTED The API is not supported. + * - \ref NVML_ERROR_IN_USE The range was not overridden because a VM is running on the host. + * - \ref NVML_ERROR_INVALID_ARGUMENT The \a vgpuVersion parameter specifies a range that is outside the range supported by the NVIDIA vGPU Manager or if \a vgpuVersion is NULL. + */ +nvmlReturn_t DECLDIR nvmlSetVgpuVersion(nvmlVgpuVersion_t *vgpuVersion); + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlUtil vGPU Utilization and Accounting + * This chapter describes operations that are associated with vGPU Utilization and Accounting. + * @{ + */ +/***************************************************************************************************/ + /** * Retrieves current utilization for vGPUs on a physical GPU (device). * @@ -5558,123 +8821,1535 @@ nvmlReturn_t DECLDIR nvmlDeviceGetVgpuProcessUtilization(nvmlDevice_t device, un unsigned int *vgpuProcessSamplesCount, nvmlVgpuProcessUtilizationSample_t *utilizationSamples); /** - * Retrieve the GRID licensable features. + * Queries the state of per process accounting mode on vGPU. * - * Identifies whether the system supports GRID Software Licensing. If it does, return the list of licensable feature(s) - * and their current license status. + * For Maxwell &tm; or newer fully supported devices. * - * @param device Identifier of the target device - * @param pGridLicensableFeatures Pointer to structure in which GRID licensable features are returned + * @param vgpuInstance The identifier of the target vGPU instance + * @param mode Reference in which to return the current accounting mode * * @return - * - \ref NVML_SUCCESS if licensable features are successfully retrieved - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a pGridLicensableFeatures is NULL + * - \ref NVML_SUCCESS if the mode has been successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a mode is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_NOT_SUPPORTED if the vGPU doesn't support this feature + * - \ref NVML_ERROR_DRIVER_NOT_LOADED if NVIDIA driver is not running on the vGPU instance * - \ref NVML_ERROR_UNKNOWN on any unexpected error */ -nvmlReturn_t DECLDIR nvmlDeviceGetGridLicensableFeatures(nvmlDevice_t device, nvmlGridLicensableFeatures_t *pGridLicensableFeatures); +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetAccountingMode(nvmlVgpuInstance_t vgpuInstance, nvmlEnableState_t *mode); /** - * Retrieves the current encoder statistics of a vGPU Instance + * Queries list of processes running on vGPU that can be queried for accounting stats. The list of processes + * returned can be in running or terminated state. * * For Maxwell &tm; or newer fully supported devices. * - * @param vgpuInstance Identifier of the target vGPU instance - * @param sessionCount Reference to an unsigned int for count of active encoder sessions - * @param averageFps Reference to an unsigned int for trailing average FPS of all active sessions - * @param averageLatency Reference to an unsigned int for encode latency in microseconds + * To just query the maximum number of processes that can be queried, call this function with *count = 0 and + * pids=NULL. The return code will be NVML_ERROR_INSUFFICIENT_SIZE, or NVML_SUCCESS if list is empty. + * + * For more details see \ref nvmlVgpuInstanceGetAccountingStats. + * + * @note In case of PID collision some processes might not be accessible before the circular buffer is full. + * + * @param vgpuInstance The identifier of the target vGPU instance + * @param count Reference in which to provide the \a pids array size, and + * to return the number of elements ready to be queried + * @param pids Reference in which to return list of process ids * * @return - * - \ref NVML_SUCCESS if \a sessionCount, \a averageFps and \a averageLatency is fetched - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a sessionCount , or \a averageFps or \a averageLatency is NULL - * or \a vgpuInstance is invalid. - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlVgpuInstanceGetEncoderStats(nvmlVgpuInstance_t vgpuInstance, unsigned int *sessionCount, - unsigned int *averageFps, unsigned int *averageLatency); - -/** - * Retrieves information about all active encoder sessions on a vGPU Instance. - * - * An array of active encoder sessions is returned in the caller-supplied buffer pointed at by \a sessionInfo. The - * array elememt count is passed in \a sessionCount, and \a sessionCount is used to return the number of sessions - * written to the buffer. - * - * If the supplied buffer is not large enough to accommodate the active session array, the function returns - * NVML_ERROR_INSUFFICIENT_SIZE, with the element count of nvmlEncoderSessionInfo_t array required in \a sessionCount. - * To query the number of active encoder sessions, call this function with *sessionCount = 0. The code will return - * NVML_SUCCESS with number of active encoder sessions updated in *sessionCount. - * - * For Maxwell &tm; or newer fully supported devices. - * - * @param vgpuInstance Identifier of the target vGPU instance - * @param sessionCount Reference to caller supplied array size, and returns - * the number of sessions. - * @param sessionInfo Reference to caller supplied array in which the list - * of session information us returned. - * - * @return - * - \ref NVML_SUCCESS if \a sessionInfo is fetched - * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a sessionCount is too small, array element count is - returned in \a sessionCount - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a sessionCount is NULL or \a vgpuInstance is invalid.. - * - \ref NVML_ERROR_UNKNOWN on any unexpected error - */ -nvmlReturn_t DECLDIR nvmlVgpuInstanceGetEncoderSessions(nvmlVgpuInstance_t vgpuInstance, unsigned int *sessionCount, nvmlEncoderSessionInfo_t *sessionInfo); - -/** - * Retrieves the current utilization and process ID - * - * For Maxwell &tm; or newer fully supported devices. - * - * Reads recent utilization of GPU SM (3D/Compute), framebuffer, video encoder, and video decoder for processes running. - * Utilization values are returned as an array of utilization sample structures in the caller-supplied buffer pointed at - * by \a utilization. One utilization sample structure is returned per process running, that had some non-zero utilization - * during the last sample period. It includes the CPU timestamp at which the samples were recorded. Individual utilization values - * are returned as "unsigned int" values. - * - * To read utilization values, first determine the size of buffer required to hold the samples by invoking the function with - * \a utilization set to NULL. The caller should allocate a buffer of size - * processSamplesCount * sizeof(nvmlProcessUtilizationSample_t). Invoke the function again with the allocated buffer passed - * in \a utilization, and \a processSamplesCount set to the number of entries the buffer is sized for. - * - * On successful return, the function updates \a processSamplesCount with the number of process utilization sample - * structures that were actually written. This may differ from a previously read value as instances are created or - * destroyed. - * - * lastSeenTimeStamp represents the CPU timestamp in microseconds at which utilization samples were last read. Set it to 0 - * to read utilization based on all the samples maintained by the driver's internal sample buffer. Set lastSeenTimeStamp - * to a timeStamp retrieved from a previous query to read utilization since the previous query. - * - * @param device The identifier of the target device - * @param utilization Pointer to caller-supplied buffer in which guest process utilization samples are returned - * @param processSamplesCount Pointer to caller-supplied array size, and returns number of processes running - * @param lastSeenTimeStamp Return only samples with timestamp greater than lastSeenTimeStamp. - - * @return - * - \ref NVML_SUCCESS if \a utilization has been populated + * - \ref NVML_SUCCESS if pids were successfully retrieved * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized - * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid, \a utilization is NULL, or \a samplingPeriodUs is NULL + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a count is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_NOT_SUPPORTED if the vGPU doesn't support this feature or accounting mode is disabled + * - \ref NVML_ERROR_INSUFFICIENT_SIZE if \a count is too small (\a count is set to expected value) + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see nvmlVgpuInstanceGetAccountingPids + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetAccountingPids(nvmlVgpuInstance_t vgpuInstance, unsigned int *count, unsigned int *pids); + +/** + * Queries process's accounting stats. + * + * For Maxwell &tm; or newer fully supported devices. + * + * Accounting stats capture GPU utilization and other statistics across the lifetime of a process, and + * can be queried during life time of the process or after its termination. + * The time field in \ref nvmlAccountingStats_t is reported as 0 during the lifetime of the process and + * updated to actual running time after its termination. + * Accounting stats are kept in a circular buffer, newly created processes overwrite information about old + * processes. + * + * See \ref nvmlAccountingStats_t for description of each returned metric. + * List of processes that can be queried can be retrieved from \ref nvmlVgpuInstanceGetAccountingPids. + * + * @note Accounting Mode needs to be on. See \ref nvmlVgpuInstanceGetAccountingMode. + * @note Only compute and graphics applications stats can be queried. Monitoring applications stats can't be + * queried since they don't contribute to GPU utilization. + * @note In case of pid collision stats of only the latest process (that terminated last) will be reported + * + * @param vgpuInstance The identifier of the target vGPU instance + * @param pid Process Id of the target process to query stats for + * @param stats Reference in which to return the process's accounting stats + * + * @return + * - \ref NVML_SUCCESS if stats have been successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a stats is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * or \a stats is not found + * - \ref NVML_ERROR_NOT_SUPPORTED if the vGPU doesn't support this feature or accounting mode is disabled + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetAccountingStats(nvmlVgpuInstance_t vgpuInstance, unsigned int pid, nvmlAccountingStats_t *stats); + +/** + * Clears accounting information of the vGPU instance that have already terminated. + * + * For Maxwell &tm; or newer fully supported devices. + * Requires root/admin permissions. + * + * @note Accounting Mode needs to be on. See \ref nvmlVgpuInstanceGetAccountingMode. + * @note Only compute and graphics applications stats are reported and can be cleared since monitoring applications + * stats don't contribute to GPU utilization. + * + * @param vgpuInstance The identifier of the target vGPU instance + * + * @return + * - \ref NVML_SUCCESS if accounting information has been cleared + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is invalid + * - \ref NVML_ERROR_NO_PERMISSION if the user doesn't have permission to perform this operation + * - \ref NVML_ERROR_NOT_SUPPORTED if the vGPU doesn't support this feature or accounting mode is disabled + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceClearAccountingPids(nvmlVgpuInstance_t vgpuInstance); + +/** + * Query the license information of the vGPU instance. + * + * For Maxwell &tm; or newer fully supported devices. + * + * @param vgpuInstance Identifier of the target vGPU instance + * @param licenseInfo Pointer to vGPU license information structure + * + * @return + * - \ref NVML_SUCCESS if information is successfully retrieved + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a vgpuInstance is 0, or \a licenseInfo is NULL + * - \ref NVML_ERROR_NOT_FOUND if \a vgpuInstance does not match a valid active vGPU instance on the system + * - \ref NVML_ERROR_DRIVER_NOT_LOADED if NVIDIA driver is not running on the vGPU instance + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetLicenseInfo_v2(nvmlVgpuInstance_t vgpuInstance, nvmlVgpuLicenseInfo_t *licenseInfo); +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlExcludedGpuQueries Excluded GPU Queries + * This chapter describes NVML operations that are associated with excluded GPUs. + * @{ + */ +/***************************************************************************************************/ + +/** + * Excluded GPU device information + **/ +typedef struct nvmlExcludedDeviceInfo_st +{ + nvmlPciInfo_t pciInfo; //!< The PCI information for the excluded GPU + char uuid[NVML_DEVICE_UUID_BUFFER_SIZE]; //!< The ASCII string UUID for the excluded GPU +} nvmlExcludedDeviceInfo_t; + + /** + * Retrieves the number of excluded GPU devices in the system. + * + * For all products. + * + * @param deviceCount Reference in which to return the number of excluded devices + * + * @return + * - \ref NVML_SUCCESS if \a deviceCount has been set + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a deviceCount is NULL + */ +nvmlReturn_t DECLDIR nvmlGetExcludedDeviceCount(unsigned int *deviceCount); + +/** + * Acquire the device information for an excluded GPU device, based on its index. + * + * For all products. + * + * Valid indices are derived from the \a deviceCount returned by + * \ref nvmlGetExcludedDeviceCount(). For example, if \a deviceCount is 2 the valid indices + * are 0 and 1, corresponding to GPU 0 and GPU 1. + * + * @param index The index of the target GPU, >= 0 and < \a deviceCount + * @param info Reference in which to return the device information + * + * @return + * - \ref NVML_SUCCESS if \a device has been set + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a index is invalid or \a info is NULL + * + * @see nvmlGetExcludedDeviceCount + */ +nvmlReturn_t DECLDIR nvmlGetExcludedDeviceInfoByIndex(unsigned int index, nvmlExcludedDeviceInfo_t *info); + +/** @} */ + +/***************************************************************************************************/ +/** @defgroup nvmlMultiInstanceGPU Multi Instance GPU Management + * This chapter describes NVML operations that are associated with Multi Instance GPU management. + * @{ + */ +/***************************************************************************************************/ + +/** + * Disable Multi Instance GPU mode. + */ +#define NVML_DEVICE_MIG_DISABLE 0x0 + +/** + * Enable Multi Instance GPU mode. + */ +#define NVML_DEVICE_MIG_ENABLE 0x1 + +/** + * GPU instance profiles. + * + * These macros should be passed to \ref nvmlDeviceGetGpuInstanceProfileInfo to retrieve the + * detailed information about a GPU instance such as profile ID, engine counts. + */ +#define NVML_GPU_INSTANCE_PROFILE_1_SLICE 0x0 +#define NVML_GPU_INSTANCE_PROFILE_2_SLICE 0x1 +#define NVML_GPU_INSTANCE_PROFILE_3_SLICE 0x2 +#define NVML_GPU_INSTANCE_PROFILE_4_SLICE 0x3 +#define NVML_GPU_INSTANCE_PROFILE_7_SLICE 0x4 +#define NVML_GPU_INSTANCE_PROFILE_8_SLICE 0x5 +#define NVML_GPU_INSTANCE_PROFILE_6_SLICE 0x6 +#define NVML_GPU_INSTANCE_PROFILE_1_SLICE_REV1 0x7 +#define NVML_GPU_INSTANCE_PROFILE_2_SLICE_REV1 0x8 +#define NVML_GPU_INSTANCE_PROFILE_1_SLICE_REV2 0x9 +#define NVML_GPU_INSTANCE_PROFILE_COUNT 0xA + +/** + * MIG GPU instance profile capability. + * + * Bit field values representing MIG profile capabilities + * \ref nvmlGpuInstanceProfileInfo_v3_t.capabilities + */ +#define NVML_GPU_INTSTANCE_PROFILE_CAPS_P2P 0x1 + +/** + * MIG compute instance profile capability. + * + * Bit field values representing MIG profile capabilities + * \ref nvmlComputeInstanceProfileInfo_v3_t.capabilities + */ +/* No capabilities for compute profiles currently exposed */ + +typedef struct nvmlGpuInstancePlacement_st +{ + unsigned int start; //!< Index of first occupied memory slice + unsigned int size; //!< Number of memory slices occupied +} nvmlGpuInstancePlacement_t; + +/** + * GPU instance profile information. + */ +typedef struct nvmlGpuInstanceProfileInfo_st +{ + unsigned int id; //!< Unique profile ID within the device + unsigned int isP2pSupported; //!< Peer-to-Peer support + unsigned int sliceCount; //!< GPU Slice count + unsigned int instanceCount; //!< GPU instance count + unsigned int multiprocessorCount; //!< Streaming Multiprocessor count + unsigned int copyEngineCount; //!< Copy Engine count + unsigned int decoderCount; //!< Decoder Engine count + unsigned int encoderCount; //!< Encoder Engine count + unsigned int jpegCount; //!< JPEG Engine count + unsigned int ofaCount; //!< OFA Engine count + unsigned long long memorySizeMB; //!< Memory size in MBytes +} nvmlGpuInstanceProfileInfo_t; + +/** + * GPU instance profile information (v2). + * + * Version 2 adds the \ref nvmlGpuInstanceProfileInfo_v2_t.version field + * to the start of the structure, and the \ref nvmlGpuInstanceProfileInfo_v2_t.name + * field to the end. This structure is not backwards-compatible with + * \ref nvmlGpuInstanceProfileInfo_t. + */ +typedef struct nvmlGpuInstanceProfileInfo_v2_st +{ + unsigned int version; //!< Structure version identifier (set to \ref nvmlGpuInstanceProfileInfo_v2) + unsigned int id; //!< Unique profile ID within the device + unsigned int isP2pSupported; //!< Peer-to-Peer support + unsigned int sliceCount; //!< GPU Slice count + unsigned int instanceCount; //!< GPU instance count + unsigned int multiprocessorCount; //!< Streaming Multiprocessor count + unsigned int copyEngineCount; //!< Copy Engine count + unsigned int decoderCount; //!< Decoder Engine count + unsigned int encoderCount; //!< Encoder Engine count + unsigned int jpegCount; //!< JPEG Engine count + unsigned int ofaCount; //!< OFA Engine count + unsigned long long memorySizeMB; //!< Memory size in MBytes + char name[NVML_DEVICE_NAME_V2_BUFFER_SIZE]; //!< Profile name +} nvmlGpuInstanceProfileInfo_v2_t; + +/** + * Version identifier value for \ref nvmlGpuInstanceProfileInfo_v2_t.version. + */ +#define nvmlGpuInstanceProfileInfo_v2 NVML_STRUCT_VERSION(GpuInstanceProfileInfo, 2) + +/** + * GPU instance profile information (v3). + * + * Version 3 removes isP2pSupported field and adds the \ref nvmlGpuInstanceProfileInfo_v3_t.capabilities + * field \ref nvmlGpuInstanceProfileInfo_t. + */ +typedef struct nvmlGpuInstanceProfileInfo_v3_st +{ + unsigned int version; //!< Structure version identifier (set to \ref nvmlGpuInstanceProfileInfo_v3) + unsigned int id; //!< Unique profile ID within the device + unsigned int sliceCount; //!< GPU Slice count + unsigned int instanceCount; //!< GPU instance count + unsigned int multiprocessorCount; //!< Streaming Multiprocessor count + unsigned int copyEngineCount; //!< Copy Engine count + unsigned int decoderCount; //!< Decoder Engine count + unsigned int encoderCount; //!< Encoder Engine count + unsigned int jpegCount; //!< JPEG Engine count + unsigned int ofaCount; //!< OFA Engine count + unsigned long long memorySizeMB; //!< Memory size in MBytes + char name[NVML_DEVICE_NAME_V2_BUFFER_SIZE]; //!< Profile name + unsigned int capabilities; //!< Additional capabilities +} nvmlGpuInstanceProfileInfo_v3_t; + +/** + * Version identifier value for \ref nvmlGpuInstanceProfileInfo_v3_t.version. + */ +#define nvmlGpuInstanceProfileInfo_v3 NVML_STRUCT_VERSION(GpuInstanceProfileInfo, 3) + +typedef struct nvmlGpuInstanceInfo_st +{ + nvmlDevice_t device; //!< Parent device + unsigned int id; //!< Unique instance ID within the device + unsigned int profileId; //!< Unique profile ID within the device + nvmlGpuInstancePlacement_t placement; //!< Placement for this instance +} nvmlGpuInstanceInfo_t; + +typedef struct nvmlGpuInstance_st* nvmlGpuInstance_t; + +/** + * Compute instance profiles. + * + * These macros should be passed to \ref nvmlGpuInstanceGetComputeInstanceProfileInfo to retrieve the + * detailed information about a compute instance such as profile ID, engine counts + */ +#define NVML_COMPUTE_INSTANCE_PROFILE_1_SLICE 0x0 +#define NVML_COMPUTE_INSTANCE_PROFILE_2_SLICE 0x1 +#define NVML_COMPUTE_INSTANCE_PROFILE_3_SLICE 0x2 +#define NVML_COMPUTE_INSTANCE_PROFILE_4_SLICE 0x3 +#define NVML_COMPUTE_INSTANCE_PROFILE_7_SLICE 0x4 +#define NVML_COMPUTE_INSTANCE_PROFILE_8_SLICE 0x5 +#define NVML_COMPUTE_INSTANCE_PROFILE_6_SLICE 0x6 +#define NVML_COMPUTE_INSTANCE_PROFILE_1_SLICE_REV1 0x7 +#define NVML_COMPUTE_INSTANCE_PROFILE_COUNT 0x8 + +#define NVML_COMPUTE_INSTANCE_ENGINE_PROFILE_SHARED 0x0 //!< All the engines except multiprocessors would be shared +#define NVML_COMPUTE_INSTANCE_ENGINE_PROFILE_COUNT 0x1 + +typedef struct nvmlComputeInstancePlacement_st +{ + unsigned int start; //!< Index of first occupied compute slice + unsigned int size; //!< Number of compute slices occupied +} nvmlComputeInstancePlacement_t; + +/** + * Compute instance profile information. + */ +typedef struct nvmlComputeInstanceProfileInfo_st +{ + unsigned int id; //!< Unique profile ID within the GPU instance + unsigned int sliceCount; //!< GPU Slice count + unsigned int instanceCount; //!< Compute instance count + unsigned int multiprocessorCount; //!< Streaming Multiprocessor count + unsigned int sharedCopyEngineCount; //!< Shared Copy Engine count + unsigned int sharedDecoderCount; //!< Shared Decoder Engine count + unsigned int sharedEncoderCount; //!< Shared Encoder Engine count + unsigned int sharedJpegCount; //!< Shared JPEG Engine count + unsigned int sharedOfaCount; //!< Shared OFA Engine count +} nvmlComputeInstanceProfileInfo_t; + +/** + * Compute instance profile information (v2). + * + * Version 2 adds the \ref nvmlComputeInstanceProfileInfo_v2_t.version field + * to the start of the structure, and the \ref nvmlComputeInstanceProfileInfo_v2_t.name + * field to the end. This structure is not backwards-compatible with + * \ref nvmlComputeInstanceProfileInfo_t. + */ +typedef struct nvmlComputeInstanceProfileInfo_v2_st +{ + unsigned int version; //!< Structure version identifier (set to \ref nvmlComputeInstanceProfileInfo_v2) + unsigned int id; //!< Unique profile ID within the GPU instance + unsigned int sliceCount; //!< GPU Slice count + unsigned int instanceCount; //!< Compute instance count + unsigned int multiprocessorCount; //!< Streaming Multiprocessor count + unsigned int sharedCopyEngineCount; //!< Shared Copy Engine count + unsigned int sharedDecoderCount; //!< Shared Decoder Engine count + unsigned int sharedEncoderCount; //!< Shared Encoder Engine count + unsigned int sharedJpegCount; //!< Shared JPEG Engine count + unsigned int sharedOfaCount; //!< Shared OFA Engine count + char name[NVML_DEVICE_NAME_V2_BUFFER_SIZE]; //!< Profile name +} nvmlComputeInstanceProfileInfo_v2_t; + +/** + * Version identifier value for \ref nvmlComputeInstanceProfileInfo_v2_t.version. + */ +#define nvmlComputeInstanceProfileInfo_v2 NVML_STRUCT_VERSION(ComputeInstanceProfileInfo, 2) + +/** + * Compute instance profile information (v3). + * + * Version 3 adds the \ref nvmlComputeInstanceProfileInfo_v3_t.capabilities field + * \ref nvmlComputeInstanceProfileInfo_t. + */ +typedef struct nvmlComputeInstanceProfileInfo_v3_st +{ + unsigned int version; //!< Structure version identifier (set to \ref nvmlComputeInstanceProfileInfo_v3) + unsigned int id; //!< Unique profile ID within the GPU instance + unsigned int sliceCount; //!< GPU Slice count + unsigned int instanceCount; //!< Compute instance count + unsigned int multiprocessorCount; //!< Streaming Multiprocessor count + unsigned int sharedCopyEngineCount; //!< Shared Copy Engine count + unsigned int sharedDecoderCount; //!< Shared Decoder Engine count + unsigned int sharedEncoderCount; //!< Shared Encoder Engine count + unsigned int sharedJpegCount; //!< Shared JPEG Engine count + unsigned int sharedOfaCount; //!< Shared OFA Engine count + char name[NVML_DEVICE_NAME_V2_BUFFER_SIZE]; //!< Profile name + unsigned int capabilities; //!< Additional capabilities +} nvmlComputeInstanceProfileInfo_v3_t; + +/** + * Version identifier value for \ref nvmlComputeInstanceProfileInfo_v3_t.version. + */ +#define nvmlComputeInstanceProfileInfo_v3 NVML_STRUCT_VERSION(ComputeInstanceProfileInfo, 3) + +typedef struct nvmlComputeInstanceInfo_st +{ + nvmlDevice_t device; //!< Parent device + nvmlGpuInstance_t gpuInstance; //!< Parent GPU instance + unsigned int id; //!< Unique instance ID within the GPU instance + unsigned int profileId; //!< Unique profile ID within the GPU instance + nvmlComputeInstancePlacement_t placement; //!< Placement for this instance within the GPU instance's compute slice range {0, sliceCount} +} nvmlComputeInstanceInfo_t; + +typedef struct nvmlComputeInstance_st* nvmlComputeInstance_t; + +/** + * Set MIG mode for the device. + * + * For Ampere &tm; or newer fully supported devices. + * Requires root user. + * + * This mode determines whether a GPU instance can be created. + * + * This API may unbind or reset the device to activate the requested mode. Thus, the attributes associated with the + * device, such as minor number, might change. The caller of this API is expected to query such attributes again. + * + * On certain platforms like pass-through virtualization, where reset functionality may not be exposed directly, VM + * reboot is required. \a activationStatus would return \ref NVML_ERROR_RESET_REQUIRED for such cases. + * + * \a activationStatus would return the appropriate error code upon unsuccessful activation. For example, if device + * unbind fails because the device isn't idle, \ref NVML_ERROR_IN_USE would be returned. The caller of this API + * is expected to idle the device and retry setting the \a mode. + * + * @note On Windows, only disabling MIG mode is supported. \a activationStatus would return \ref + * NVML_ERROR_NOT_SUPPORTED as GPU reset is not supported on Windows through this API. + * + * @param device The identifier of the target device + * @param mode The mode to be set, \ref NVML_DEVICE_MIG_DISABLE or + * \ref NVML_DEVICE_MIG_ENABLE + * @param activationStatus The activationStatus status + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device,\a mode or \a activationStatus are invalid + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't support MIG mode + */ +nvmlReturn_t DECLDIR nvmlDeviceSetMigMode(nvmlDevice_t device, unsigned int mode, nvmlReturn_t *activationStatus); + +/** + * Get MIG mode for the device. + * + * For Ampere &tm; or newer fully supported devices. + * + * Changing MIG modes may require device unbind or reset. The "pending" MIG mode refers to the target mode following the + * next activation trigger. + * + * @param device The identifier of the target device + * @param currentMode Returns the current mode, \ref NVML_DEVICE_MIG_DISABLE or + * \ref NVML_DEVICE_MIG_ENABLE + * @param pendingMode Returns the pending mode, \ref NVML_DEVICE_MIG_DISABLE or + * \ref NVML_DEVICE_MIG_ENABLE + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a currentMode or \a pendingMode are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't support MIG mode + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMigMode(nvmlDevice_t device, unsigned int *currentMode, unsigned int *pendingMode); + +/** + * Get GPU instance profile information. + * + * Information provided by this API is immutable throughout the lifetime of a MIG mode. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device The identifier of the target device + * @param profile One of the NVML_GPU_INSTANCE_PROFILE_* + * @param info Returns detailed profile information + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a profile or \a info are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled or \a profile isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuInstanceProfileInfo(nvmlDevice_t device, unsigned int profile, + nvmlGpuInstanceProfileInfo_t *info); + +/** + * Versioned wrapper around \ref nvmlDeviceGetGpuInstanceProfileInfo that accepts a versioned + * \ref nvmlGpuInstanceProfileInfo_v2_t or later output structure. + * + * @note The caller must set the \ref nvmlGpuInstanceProfileInfo_v2_t.version field to the + * appropriate version prior to calling this function. For example: + * \code + * nvmlGpuInstanceProfileInfo_v2_t profileInfo = + * { .version = nvmlGpuInstanceProfileInfo_v2 }; + * nvmlReturn_t result = nvmlDeviceGetGpuInstanceProfileInfoV(device, + * profile, + * &profileInfo); + * \endcode + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device The identifier of the target device + * @param profile One of the NVML_GPU_INSTANCE_PROFILE_* + * @param info Returns detailed profile information + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a profile, \a info, or \a info->version are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled or \a profile isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuInstanceProfileInfoV(nvmlDevice_t device, unsigned int profile, + nvmlGpuInstanceProfileInfo_v2_t *info); + +/** + * Get GPU instance placements. + * + * A placement represents the location of a GPU instance within a device. This API only returns all the possible + * placements for the given profile. + * A created GPU instance occupies memory slices described by its placement. Creation of new GPU instance will + * fail if there is overlap with the already occupied memory slices. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param device The identifier of the target device + * @param profileId The GPU instance profile ID. See \ref nvmlDeviceGetGpuInstanceProfileInfo + * @param placements Returns placements allowed for the profile. Can be NULL to discover number + * of allowed placements for this profile. If non-NULL must be large enough + * to accommodate the placements supported by the profile. + * @param count Returns number of allowed placemenets for the profile. + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a profileId or \a count are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled or \a profileId isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuInstancePossiblePlacements_v2(nvmlDevice_t device, unsigned int profileId, + nvmlGpuInstancePlacement_t *placements, + unsigned int *count); + +/** + * Get GPU instance profile capacity. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param device The identifier of the target device + * @param profileId The GPU instance profile ID. See \ref nvmlDeviceGetGpuInstanceProfileInfo + * @param count Returns remaining instance count for the profile ID + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a profileId or \a count are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled or \a profileId isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuInstanceRemainingCapacity(nvmlDevice_t device, unsigned int profileId, + unsigned int *count); + +/** + * Create GPU instance. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * If the parent device is unbound, reset or the GPU instance is destroyed explicitly, the GPU instance handle would + * become invalid. The GPU instance must be recreated to acquire a valid handle. + * + * @param device The identifier of the target device + * @param profileId The GPU instance profile ID. See \ref nvmlDeviceGetGpuInstanceProfileInfo + * @param gpuInstance Returns the GPU instance handle + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a profile, \a profileId or \a gpuInstance are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled or in vGPU guest + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_INSUFFICIENT_RESOURCES If the requested GPU instance could not be created + */ +nvmlReturn_t DECLDIR nvmlDeviceCreateGpuInstance(nvmlDevice_t device, unsigned int profileId, + nvmlGpuInstance_t *gpuInstance); + +/** + * Create GPU instance with the specified placement. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * If the parent device is unbound, reset or the GPU instance is destroyed explicitly, the GPU instance handle would + * become invalid. The GPU instance must be recreated to acquire a valid handle. + * + * @param device The identifier of the target device + * @param profileId The GPU instance profile ID. See \ref nvmlDeviceGetGpuInstanceProfileInfo + * @param placement The requested placement. See \ref nvmlDeviceGetGpuInstancePossiblePlacements_v2 + * @param gpuInstance Returns the GPU instance handle + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a profile, \a profileId, \a placement or \a gpuInstance + * are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled or in vGPU guest + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_INSUFFICIENT_RESOURCES If the requested GPU instance could not be created + */ +nvmlReturn_t DECLDIR nvmlDeviceCreateGpuInstanceWithPlacement(nvmlDevice_t device, unsigned int profileId, + const nvmlGpuInstancePlacement_t *placement, + nvmlGpuInstance_t *gpuInstance); +/** + * Destroy GPU instance. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param gpuInstance The GPU instance handle + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled or in vGPU guest + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_IN_USE If the GPU instance is in use. This error would be returned if processes + * (e.g. CUDA application) or compute instances are active on the + * GPU instance. + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceDestroy(nvmlGpuInstance_t gpuInstance); + +/** + * Get GPU instances for given profile ID. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param device The identifier of the target device + * @param profileId The GPU instance profile ID. See \ref nvmlDeviceGetGpuInstanceProfileInfo + * @param gpuInstances Returns pre-exiting GPU instances, the buffer must be large enough to + * accommodate the instances supported by the profile. + * See \ref nvmlDeviceGetGpuInstanceProfileInfo + * @param count The count of returned GPU instances + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a profileId, \a gpuInstances or \a count are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuInstances(nvmlDevice_t device, unsigned int profileId, + nvmlGpuInstance_t *gpuInstances, unsigned int *count); + +/** + * Get GPU instances for given instance ID. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param device The identifier of the target device + * @param id The GPU instance ID + * @param gpuInstance Returns GPU instance + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a id or \a gpuInstance are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_NOT_FOUND If the GPU instance is not found. + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuInstanceById(nvmlDevice_t device, unsigned int id, nvmlGpuInstance_t *gpuInstance); + +/** + * Get GPU instance information. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param gpuInstance The GPU instance handle + * @param info Return GPU instance information + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance or \a info are invalid + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetInfo(nvmlGpuInstance_t gpuInstance, nvmlGpuInstanceInfo_t *info); + +/** + * Get compute instance profile information. + * + * Information provided by this API is immutable throughout the lifetime of a MIG mode. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param gpuInstance The identifier of the target GPU instance + * @param profile One of the NVML_COMPUTE_INSTANCE_PROFILE_* + * @param engProfile One of the NVML_COMPUTE_INSTANCE_ENGINE_PROFILE_* + * @param info Returns detailed profile information + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance, \a profile, \a engProfile or \a info are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a profile isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetComputeInstanceProfileInfo(nvmlGpuInstance_t gpuInstance, unsigned int profile, + unsigned int engProfile, + nvmlComputeInstanceProfileInfo_t *info); + +/** + * Versioned wrapper around \ref nvmlGpuInstanceGetComputeInstanceProfileInfo that accepts a versioned + * \ref nvmlComputeInstanceProfileInfo_v2_t or later output structure. + * + * @note The caller must set the \ref nvmlGpuInstanceProfileInfo_v2_t.version field to the + * appropriate version prior to calling this function. For example: + * \code + * nvmlComputeInstanceProfileInfo_v2_t profileInfo = + * { .version = nvmlComputeInstanceProfileInfo_v2 }; + * nvmlReturn_t result = nvmlGpuInstanceGetComputeInstanceProfileInfoV(gpuInstance, + * profile, + * engProfile, + * &profileInfo); + * \endcode + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param gpuInstance The identifier of the target GPU instance + * @param profile One of the NVML_COMPUTE_INSTANCE_PROFILE_* + * @param engProfile One of the NVML_COMPUTE_INSTANCE_ENGINE_PROFILE_* + * @param info Returns detailed profile information + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance, \a profile, \a engProfile, \a info, or \a info->version are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a profile isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetComputeInstanceProfileInfoV(nvmlGpuInstance_t gpuInstance, unsigned int profile, + unsigned int engProfile, + nvmlComputeInstanceProfileInfo_v2_t *info); + +/** + * Get compute instance profile capacity. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param gpuInstance The identifier of the target GPU instance + * @param profileId The compute instance profile ID. + * See \ref nvmlGpuInstanceGetComputeInstanceProfileInfo + * @param count Returns remaining instance count for the profile ID + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance, \a profileId or \a availableCount are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a profileId isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetComputeInstanceRemainingCapacity(nvmlGpuInstance_t gpuInstance, + unsigned int profileId, unsigned int *count); + +/** + * Get compute instance placements. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * A placement represents the location of a compute instance within a GPU instance. This API only returns all the possible + * placements for the given profile. + * A created compute instance occupies compute slices described by its placement. Creation of new compute instance will + * fail if there is overlap with the already occupied compute slices. + * + * @param gpuInstance The identifier of the target GPU instance + * @param profileId The compute instance profile ID. See \ref nvmlGpuInstanceGetComputeInstanceProfileInfo + * @param placements Returns placements allowed for the profile. Can be NULL to discover number + * of allowed placements for this profile. If non-NULL must be large enough + * to accommodate the placements supported by the profile. + * @param count Returns number of allowed placemenets for the profile. + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance, \a profileId or \a count are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled or \a profileId isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetComputeInstancePossiblePlacements(nvmlGpuInstance_t gpuInstance, + unsigned int profileId, + nvmlComputeInstancePlacement_t *placements, + unsigned int *count); + +/** + * Create compute instance. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * If the parent device is unbound, reset or the parent GPU instance is destroyed or the compute instance is destroyed + * explicitly, the compute instance handle would become invalid. The compute instance must be recreated to acquire + * a valid handle. + * + * @param gpuInstance The identifier of the target GPU instance + * @param profileId The compute instance profile ID. + * See \ref nvmlGpuInstanceGetComputeInstanceProfileInfo + * @param computeInstance Returns the compute instance handle + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance, \a profile, \a profileId or \a computeInstance + * are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a profileId isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_INSUFFICIENT_RESOURCES If the requested compute instance could not be created + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceCreateComputeInstance(nvmlGpuInstance_t gpuInstance, unsigned int profileId, + nvmlComputeInstance_t *computeInstance); + +/** + * Create compute instance with the specified placement. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * If the parent device is unbound, reset or the parent GPU instance is destroyed or the compute instance is destroyed + * explicitly, the compute instance handle would become invalid. The compute instance must be recreated to acquire + * a valid handle. + * + * @param gpuInstance The identifier of the target GPU instance + * @param profileId The compute instance profile ID. + * See \ref nvmlGpuInstanceGetComputeInstanceProfileInfo + * @param placement The requested placement. See \ref nvmlGpuInstanceGetComputeInstancePossiblePlacements + * @param computeInstance Returns the compute instance handle + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance, \a profile, \a profileId or \a computeInstance + * are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a profileId isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_INSUFFICIENT_RESOURCES If the requested compute instance could not be created + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceCreateComputeInstanceWithPlacement(nvmlGpuInstance_t gpuInstance, unsigned int profileId, + const nvmlComputeInstancePlacement_t *placement, + nvmlComputeInstance_t *computeInstance); + +/** + * Destroy compute instance. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param computeInstance The compute instance handle + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a computeInstance is invalid + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_IN_USE If the compute instance is in use. This error would be returned if + * processes (e.g. CUDA application) are active on the compute instance. + */ +nvmlReturn_t DECLDIR nvmlComputeInstanceDestroy(nvmlComputeInstance_t computeInstance); + +/** + * Get compute instances for given profile ID. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param gpuInstance The identifier of the target GPU instance + * @param profileId The compute instance profile ID. + * See \ref nvmlGpuInstanceGetComputeInstanceProfileInfo + * @param computeInstances Returns pre-exiting compute instances, the buffer must be large enough to + * accommodate the instances supported by the profile. + * See \ref nvmlGpuInstanceGetComputeInstanceProfileInfo + * @param count The count of returned compute instances + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a gpuInstance, \a profileId, \a computeInstances or \a count + * are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a profileId isn't supported + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetComputeInstances(nvmlGpuInstance_t gpuInstance, unsigned int profileId, + nvmlComputeInstance_t *computeInstances, unsigned int *count); + +/** + * Get compute instance for given instance ID. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * Requires privileged user. + * + * @param gpuInstance The identifier of the target GPU instance + * @param id The compute instance ID + * @param computeInstance Returns compute instance + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a device, \a ID or \a computeInstance are invalid + * - \ref NVML_ERROR_NOT_SUPPORTED If \a device doesn't have MIG mode enabled + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + * - \ref NVML_ERROR_NOT_FOUND If the compute instance is not found. + */ +nvmlReturn_t DECLDIR nvmlGpuInstanceGetComputeInstanceById(nvmlGpuInstance_t gpuInstance, unsigned int id, + nvmlComputeInstance_t *computeInstance); + +/** + * Get compute instance information. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param computeInstance The compute instance handle + * @param info Return compute instance information + * + * @return + * - \ref NVML_SUCCESS Upon success + * - \ref NVML_ERROR_UNINITIALIZED If library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT If \a computeInstance or \a info are invalid + * - \ref NVML_ERROR_NO_PERMISSION If user doesn't have permission to perform the operation + */ +nvmlReturn_t DECLDIR nvmlComputeInstanceGetInfo_v2(nvmlComputeInstance_t computeInstance, nvmlComputeInstanceInfo_t *info); + +/** + * Test if the given handle refers to a MIG device. + * + * A MIG device handle is an NVML abstraction which maps to a MIG compute instance. + * These overloaded references can be used (with some restrictions) interchangeably + * with a GPU device handle to execute queries at a per-compute instance granularity. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device NVML handle to test + * @param isMigDevice True when handle refers to a MIG device + * + * @return + * - \ref NVML_SUCCESS if \a device status was successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device handle or \a isMigDevice reference is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this check is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceIsMigDeviceHandle(nvmlDevice_t device, unsigned int *isMigDevice); + +/** + * Get GPU instance ID for the given MIG device handle. + * + * GPU instance IDs are unique per device and remain valid until the GPU instance is destroyed. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device Target MIG device handle + * @param id GPU instance ID + * + * @return + * - \ref NVML_SUCCESS if instance ID was successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a id reference is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetGpuInstanceId(nvmlDevice_t device, unsigned int *id); + +/** + * Get compute instance ID for the given MIG device handle. + * + * Compute instance IDs are unique per GPU instance and remain valid until the compute instance + * is destroyed. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device Target MIG device handle + * @param id Compute instance ID + * + * @return + * - \ref NVML_SUCCESS if instance ID was successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a id reference is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetComputeInstanceId(nvmlDevice_t device, unsigned int *id); + +/** + * Get the maximum number of MIG devices that can exist under a given parent NVML device. + * + * Returns zero if MIG is not supported or enabled. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device Target device handle + * @param count Count of MIG devices + * + * @return + * - \ref NVML_SUCCESS if \a count was successfully retrieved + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device or \a count reference is invalid + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMaxMigDeviceCount(nvmlDevice_t device, unsigned int *count); + +/** + * Get MIG device handle for the given index under its parent NVML device. + * + * If the compute instance is destroyed either explicitly or by destroying, + * resetting or unbinding the parent GPU instance or the GPU device itself + * the MIG device handle would remain invalid and must be requested again + * using this API. Handles may be reused and their properties can change in + * the process. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param device Reference to the parent GPU device handle + * @param index Index of the MIG device + * @param migDevice Reference to the MIG device handle + * + * @return + * - \ref NVML_SUCCESS if \a migDevice handle was successfully created + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device, \a index or \a migDevice reference is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_NOT_FOUND if no valid MIG device was found at \a index + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetMigDeviceHandleByIndex(nvmlDevice_t device, unsigned int index, + nvmlDevice_t *migDevice); + +/** + * Get parent device handle from a MIG device handle. + * + * For Ampere &tm; or newer fully supported devices. + * Supported on Linux only. + * + * @param migDevice MIG device handle + * @param device Device handle + * + * @return + * - \ref NVML_SUCCESS if \a device handle was successfully created + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a migDevice or \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * - \ref NVML_ERROR_UNKNOWN on any unexpected error + */ +nvmlReturn_t DECLDIR nvmlDeviceGetDeviceHandleFromMigDeviceHandle(nvmlDevice_t migDevice, nvmlDevice_t *device); + +/** @} */ // @defgroup nvmlMultiInstanceGPU + + +/***************************************************************************************************/ +/** @defgroup GPM NVML GPM + * @{ + */ +/***************************************************************************************************/ +/** @defgroup nvmlGpmEnums GPM Enums + * @{ + */ +/***************************************************************************************************/ + +/** + * GPM Metric Identifiers + */ +typedef enum +{ + NVML_GPM_METRIC_GRAPHICS_UTIL = 1, //!< Percentage of time any compute/graphics app was active on the GPU. 0.0 - 100.0 + NVML_GPM_METRIC_SM_UTIL = 2, //!< Percentage of SMs that were busy. 0.0 - 100.0 + NVML_GPM_METRIC_SM_OCCUPANCY = 3, //!< Percentage of warps that were active vs theoretical maximum. 0.0 - 100.0 + NVML_GPM_METRIC_INTEGER_UTIL = 4, //!< Percentage of time the GPU's SMs were doing integer operations. 0.0 - 100.0 + NVML_GPM_METRIC_ANY_TENSOR_UTIL = 5, //!< Percentage of time the GPU's SMs were doing ANY tensor operations. 0.0 - 100.0 + NVML_GPM_METRIC_DFMA_TENSOR_UTIL = 6, //!< Percentage of time the GPU's SMs were doing DFMA tensor operations. 0.0 - 100.0 + NVML_GPM_METRIC_HMMA_TENSOR_UTIL = 7, //!< Percentage of time the GPU's SMs were doing HMMA tensor operations. 0.0 - 100.0 + NVML_GPM_METRIC_IMMA_TENSOR_UTIL = 9, //!< Percentage of time the GPU's SMs were doing IMMA tensor operations. 0.0 - 100.0 + NVML_GPM_METRIC_DRAM_BW_UTIL = 10, //!< Percentage of DRAM bw used vs theoretical maximum. 0.0 - 100.0 */ + NVML_GPM_METRIC_FP64_UTIL = 11, //!< Percentage of time the GPU's SMs were doing non-tensor FP64 math. 0.0 - 100.0 + NVML_GPM_METRIC_FP32_UTIL = 12, //!< Percentage of time the GPU's SMs were doing non-tensor FP32 math. 0.0 - 100.0 + NVML_GPM_METRIC_FP16_UTIL = 13, //!< Percentage of time the GPU's SMs were doing non-tensor FP16 math. 0.0 - 100.0 + NVML_GPM_METRIC_PCIE_TX_PER_SEC = 20, //!< PCIe traffic from this GPU in MiB/sec + NVML_GPM_METRIC_PCIE_RX_PER_SEC = 21, //!< PCIe traffic to this GPU in MiB/sec + NVML_GPM_METRIC_NVDEC_0_UTIL = 30, //!< Percent utilization of NVDEC 0. 0.0 - 100.0 + NVML_GPM_METRIC_NVDEC_1_UTIL = 31, //!< Percent utilization of NVDEC 1. 0.0 - 100.0 + NVML_GPM_METRIC_NVDEC_2_UTIL = 32, //!< Percent utilization of NVDEC 2. 0.0 - 100.0 + NVML_GPM_METRIC_NVDEC_3_UTIL = 33, //!< Percent utilization of NVDEC 3. 0.0 - 100.0 + NVML_GPM_METRIC_NVDEC_4_UTIL = 34, //!< Percent utilization of NVDEC 4. 0.0 - 100.0 + NVML_GPM_METRIC_NVDEC_5_UTIL = 35, //!< Percent utilization of NVDEC 5. 0.0 - 100.0 + NVML_GPM_METRIC_NVDEC_6_UTIL = 36, //!< Percent utilization of NVDEC 6. 0.0 - 100.0 + NVML_GPM_METRIC_NVDEC_7_UTIL = 37, //!< Percent utilization of NVDEC 7. 0.0 - 100.0 + NVML_GPM_METRIC_NVJPG_0_UTIL = 40, //!< Percent utilization of NVJPG 0. 0.0 - 100.0 + NVML_GPM_METRIC_NVJPG_1_UTIL = 41, //!< Percent utilization of NVJPG 1. 0.0 - 100.0 + NVML_GPM_METRIC_NVJPG_2_UTIL = 42, //!< Percent utilization of NVJPG 2. 0.0 - 100.0 + NVML_GPM_METRIC_NVJPG_3_UTIL = 43, //!< Percent utilization of NVJPG 3. 0.0 - 100.0 + NVML_GPM_METRIC_NVJPG_4_UTIL = 44, //!< Percent utilization of NVJPG 4. 0.0 - 100.0 + NVML_GPM_METRIC_NVJPG_5_UTIL = 45, //!< Percent utilization of NVJPG 5. 0.0 - 100.0 + NVML_GPM_METRIC_NVJPG_6_UTIL = 46, //!< Percent utilization of NVJPG 6. 0.0 - 100.0 + NVML_GPM_METRIC_NVJPG_7_UTIL = 47, //!< Percent utilization of NVJPG 7. 0.0 - 100.0 + NVML_GPM_METRIC_NVOFA_0_UTIL = 50, //!< Percent utilization of NVOFA 0. 0.0 - 100.0 + NVML_GPM_METRIC_NVLINK_TOTAL_RX_PER_SEC = 60, //!< NvLink read bandwidth for all links in MiB/sec + NVML_GPM_METRIC_NVLINK_TOTAL_TX_PER_SEC = 61, //!< NvLink write bandwidth for all links in MiB/sec + NVML_GPM_METRIC_NVLINK_L0_RX_PER_SEC = 62, //!< NvLink read bandwidth for link 0 in MiB/sec + NVML_GPM_METRIC_NVLINK_L0_TX_PER_SEC = 63, //!< NvLink write bandwidth for link 0 in MiB/sec + NVML_GPM_METRIC_NVLINK_L1_RX_PER_SEC = 64, //!< NvLink read bandwidth for link 1 in MiB/sec + NVML_GPM_METRIC_NVLINK_L1_TX_PER_SEC = 65, //!< NvLink write bandwidth for link 1 in MiB/sec + NVML_GPM_METRIC_NVLINK_L2_RX_PER_SEC = 66, //!< NvLink read bandwidth for link 2 in MiB/sec + NVML_GPM_METRIC_NVLINK_L2_TX_PER_SEC = 67, //!< NvLink write bandwidth for link 2 in MiB/sec + NVML_GPM_METRIC_NVLINK_L3_RX_PER_SEC = 68, //!< NvLink read bandwidth for link 3 in MiB/sec + NVML_GPM_METRIC_NVLINK_L3_TX_PER_SEC = 69, //!< NvLink write bandwidth for link 3 in MiB/sec + NVML_GPM_METRIC_NVLINK_L4_RX_PER_SEC = 70, //!< NvLink read bandwidth for link 4 in MiB/sec + NVML_GPM_METRIC_NVLINK_L4_TX_PER_SEC = 71, //!< NvLink write bandwidth for link 4 in MiB/sec + NVML_GPM_METRIC_NVLINK_L5_RX_PER_SEC = 72, //!< NvLink read bandwidth for link 5 in MiB/sec + NVML_GPM_METRIC_NVLINK_L5_TX_PER_SEC = 73, //!< NvLink write bandwidth for link 5 in MiB/sec + NVML_GPM_METRIC_NVLINK_L6_RX_PER_SEC = 74, //!< NvLink read bandwidth for link 6 in MiB/sec + NVML_GPM_METRIC_NVLINK_L6_TX_PER_SEC = 75, //!< NvLink write bandwidth for link 6 in MiB/sec + NVML_GPM_METRIC_NVLINK_L7_RX_PER_SEC = 76, //!< NvLink read bandwidth for link 7 in MiB/sec + NVML_GPM_METRIC_NVLINK_L7_TX_PER_SEC = 77, //!< NvLink write bandwidth for link 7 in MiB/sec + NVML_GPM_METRIC_NVLINK_L8_RX_PER_SEC = 78, //!< NvLink read bandwidth for link 8 in MiB/sec + NVML_GPM_METRIC_NVLINK_L8_TX_PER_SEC = 79, //!< NvLink write bandwidth for link 8 in MiB/sec + NVML_GPM_METRIC_NVLINK_L9_RX_PER_SEC = 80, //!< NvLink read bandwidth for link 9 in MiB/sec + NVML_GPM_METRIC_NVLINK_L9_TX_PER_SEC = 81, //!< NvLink write bandwidth for link 9 in MiB/sec + NVML_GPM_METRIC_NVLINK_L10_RX_PER_SEC = 82, //!< NvLink read bandwidth for link 10 in MiB/sec + NVML_GPM_METRIC_NVLINK_L10_TX_PER_SEC = 83, //!< NvLink write bandwidth for link 10 in MiB/sec + NVML_GPM_METRIC_NVLINK_L11_RX_PER_SEC = 84, //!< NvLink read bandwidth for link 11 in MiB/sec + NVML_GPM_METRIC_NVLINK_L11_TX_PER_SEC = 85, //!< NvLink write bandwidth for link 11 in MiB/sec + NVML_GPM_METRIC_NVLINK_L12_RX_PER_SEC = 86, //!< NvLink read bandwidth for link 12 in MiB/sec + NVML_GPM_METRIC_NVLINK_L12_TX_PER_SEC = 87, //!< NvLink write bandwidth for link 12 in MiB/sec + NVML_GPM_METRIC_NVLINK_L13_RX_PER_SEC = 88, //!< NvLink read bandwidth for link 13 in MiB/sec + NVML_GPM_METRIC_NVLINK_L13_TX_PER_SEC = 89, //!< NvLink write bandwidth for link 13 in MiB/sec + NVML_GPM_METRIC_NVLINK_L14_RX_PER_SEC = 90, //!< NvLink read bandwidth for link 14 in MiB/sec + NVML_GPM_METRIC_NVLINK_L14_TX_PER_SEC = 91, //!< NvLink write bandwidth for link 14 in MiB/sec + NVML_GPM_METRIC_NVLINK_L15_RX_PER_SEC = 92, //!< NvLink read bandwidth for link 15 in MiB/sec + NVML_GPM_METRIC_NVLINK_L15_TX_PER_SEC = 93, //!< NvLink write bandwidth for link 15 in MiB/sec + NVML_GPM_METRIC_NVLINK_L16_RX_PER_SEC = 94, //!< NvLink read bandwidth for link 16 in MiB/sec + NVML_GPM_METRIC_NVLINK_L16_TX_PER_SEC = 95, //!< NvLink write bandwidth for link 16 in MiB/sec + NVML_GPM_METRIC_NVLINK_L17_RX_PER_SEC = 96, //!< NvLink read bandwidth for link 17 in MiB/sec + NVML_GPM_METRIC_NVLINK_L17_TX_PER_SEC = 97, //!< NvLink write bandwidth for link 17 in MiB/sec + NVML_GPM_METRIC_MAX = 98, //!< Maximum value above +1. Note that changing this should also change NVML_GPM_METRICS_GET_VERSION due to struct size change +} nvmlGpmMetricId_t; + +/** @} */ // @defgroup nvmlGpmEnums + + +/***************************************************************************************************/ +/** @defgroup nvmlGpmStructs GPM Structs + * @{ + */ +/***************************************************************************************************/ + +/** + * Handle to an allocated GPM sample allocated with nvmlGpmSampleAlloc(). Free this with nvmlGpmSampleFree(). + */ +typedef struct nvmlGpmSample_st* nvmlGpmSample_t; + +/** + * GPM metric information. + */ +typedef struct +{ + unsigned int metricId; //!< IN: NVML_GPM_METRIC_? #define of which metric to retrieve + nvmlReturn_t nvmlReturn; //!< OUT: Status of this metric. If this is nonzero, then value is not valid + double value; //!< OUT: Value of this metric. Is only valid if nvmlReturn is 0 (NVML_SUCCESS) + struct + { + char *shortName; + char *longName; + char *unit; + } metricInfo; //!< OUT: Metric name and unit. Those can be NULL if not defined +} nvmlGpmMetric_t; + +/** + * GPM buffer information. + */ +typedef struct +{ + unsigned int version; //!< IN: Set to NVML_GPM_METRICS_GET_VERSION + unsigned int numMetrics; //!< IN: How many metrics to retrieve in metrics[] + nvmlGpmSample_t sample1; //!< IN: Sample buffer + nvmlGpmSample_t sample2; //!< IN: Sample buffer + nvmlGpmMetric_t metrics[NVML_GPM_METRIC_MAX]; //!< IN/OUT: Array of metrics. Set metricId on call. See nvmlReturn and value on return +} nvmlGpmMetricsGet_t; + +#define NVML_GPM_METRICS_GET_VERSION 1 + +/** + * GPM device information. + */ +typedef struct +{ + unsigned int version; //!< IN: Set to NVML_GPM_SUPPORT_VERSION + unsigned int isSupportedDevice; //!< OUT: Indicates device support +} nvmlGpmSupport_t; + +#define NVML_GPM_SUPPORT_VERSION 1 + +/** @} */ // @defgroup nvmlGPMStructs + +/***************************************************************************************************/ +/** @defgroup nvmlGpmFunctions GPM Functions + * @{ + */ +/***************************************************************************************************/ + +/** + * Calculate GPM metrics from two samples. + * + * For Hopper &tm; or newer fully supported devices. + * + * @param metricsGet IN/OUT: populated \a nvmlGpmMetricsGet_t struct + * + * @return + * - \ref NVML_SUCCESS on success + * - Nonzero NVML_ERROR_? enum on error + */ +nvmlReturn_t DECLDIR nvmlGpmMetricsGet(nvmlGpmMetricsGet_t *metricsGet); + + +/** + * Free an allocated sample buffer that was allocated with \ref nvmlGpmSampleAlloc() + * + * For Hopper &tm; or newer fully supported devices. + * + * @param gpmSample Sample to free + * + * @return + * - \ref NVML_SUCCESS on success + * - \ref NVML_ERROR_INVALID_ARGUMENT if an invalid pointer is provided + */ +nvmlReturn_t DECLDIR nvmlGpmSampleFree(nvmlGpmSample_t gpmSample); + + +/** + * Allocate a sample buffer to be used with NVML GPM . You will need to allocate + * at least two of these buffers to use with the NVML GPM feature + * + * For Hopper &tm; or newer fully supported devices. + * + * @param gpmSample Where the allocated sample will be stored + * + * @return + * - \ref NVML_SUCCESS on success + * - \ref NVML_ERROR_INVALID_ARGUMENT if an invalid pointer is provided + * - \ref NVML_ERROR_MEMORY if system memory is insufficient + */ +nvmlReturn_t DECLDIR nvmlGpmSampleAlloc(nvmlGpmSample_t *gpmSample); + +/** + * Read a sample of GPM metrics into the provided \a gpmSample buffer. After + * two samples are gathered, you can call nvmlGpmMetricGet on those samples to + * retrive metrics + * + * For Hopper &tm; or newer fully supported devices. + * + * @param device Device to get samples for + * @param gpmSample Buffer to read samples into + * + * @return + * - \ref NVML_SUCCESS on success + * - Nonzero NVML_ERROR_? enum on error + */ +nvmlReturn_t DECLDIR nvmlGpmSampleGet(nvmlDevice_t device, nvmlGpmSample_t gpmSample); + +/** + * Read a sample of GPM metrics into the provided \a gpmSample buffer for a MIG GPU Instance. + * + * After two samples are gathered, you can call nvmlGpmMetricGet on those + * samples to retrive metrics + * + * For Hopper &tm; or newer fully supported devices. + * + * @param device Device to get samples for + * @param gpuInstanceId MIG GPU Instance ID + * @param gpmSample Buffer to read samples into + * + * @return + * - \ref NVML_SUCCESS on success + * - Nonzero NVML_ERROR_? enum on error + */ +nvmlReturn_t DECLDIR nvmlGpmMigSampleGet(nvmlDevice_t device, unsigned int gpuInstanceId, nvmlGpmSample_t gpmSample); + +/** + * Indicate whether the supplied device supports GPM + * + * @param device NVML device to query for + * @param gpmSupport Structure to indicate GPM support \a nvmlGpmSupport_t. Indicates + * GPM support per system for the supplied device + * + * @return + * - NVML_SUCCESS on success + * - Nonzero NVML_ERROR_? enum if there is an error in processing the query + */ +nvmlReturn_t DECLDIR nvmlGpmQueryDeviceSupport(nvmlDevice_t device, nvmlGpmSupport_t *gpmSupport); + +/* GPM Stream State */ +/** + * Get GPM stream state. + * + * %HOPPER_OR_NEWER% + * Supported on Linux, Windows TCC. + * + * @param device The identifier of the target device + * @param state Returns GPM stream state + * NVML_FEATURE_DISABLED or NVML_FEATURE_ENABLED + * + * @return + * - \ref NVML_SUCCESS if \a current GPM stream state were successfully queried + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a state is NULL + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + */ +nvmlReturn_t DECLDIR nvmlGpmQueryIfStreamingEnabled(nvmlDevice_t device, unsigned int *state); + +/** + * Set GPM stream state. + * + * %HOPPER_OR_NEWER% + * Supported on Linux, Windows TCC. + * + * @param device The identifier of the target device + * @param state GPM stream state, + * NVML_FEATURE_DISABLED or NVML_FEATURE_ENABLED + * + * @return + * - \ref NVML_SUCCESS if \a current GPM stream state is successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + */ +nvmlReturn_t DECLDIR nvmlGpmSetStreamingEnabled(nvmlDevice_t device, unsigned int state); + +/** @} */ // @defgroup nvmlGpmFunctions +/** @} */ // @defgroup GPM + +#define NVML_NVLINK_POWER_STATE_HIGH_SPEED 0x0 +#define NVML_NVLINK_POWER_STATE_LOW 0x1 + +#define NVML_NVLINK_LOW_POWER_THRESHOLD_MIN 0x1 +#define NVML_NVLINK_LOW_POWER_THRESHOLD_MAX 0x1FFF +#define NVML_NVLINK_LOW_POWER_THRESHOLD_RESET 0xFFFFFFFF + +/* Structure containing Low Power parameters */ +typedef struct nvmlNvLinkPowerThres_st +{ + unsigned int lowPwrThreshold; //!< Low power threshold (in units of 100us) +} nvmlNvLinkPowerThres_t; + +/** + * Set NvLink Low Power Threshold for device. + * + * %HOPPER_OR_NEWER% + * + * @param device The identifier of the target device + * @param info Reference to \a nvmlNvLinkPowerThres_t struct + * input parameters + * + * @return + * - \ref NVML_SUCCESS if the \a Threshold is successfully set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a Threshold is not within range + * - \ref NVML_ERROR_NOT_SUPPORTED if this query is not supported by the device + * + **/ +nvmlReturn_t DECLDIR nvmlDeviceSetNvLinkDeviceLowPowerThreshold(nvmlDevice_t device, nvmlNvLinkPowerThres_t *info); + +/** + * Set the global nvlink bandwith mode + * + * @param nvlinkBwMode nvlink bandwidth mode + * @return + * - \ref NVML_SUCCESS on success + * - \ref NVML_ERROR_INVALID_ARGUMENT if an invalid argument is provided + * - \ref NVML_ERROR_IN_USE if P2P object exists + * - \ref NVML_ERROR_NOT_SUPPORTED if GPU is not Hopper or newer architecture. + * - \ref NVML_ERROR_NO_PERMISSION if not root user + */ +nvmlReturn_t DECLDIR nvmlSystemSetNvlinkBwMode(unsigned int nvlinkBwMode); + +/** + * Get the global nvlink bandwith mode + * + * @param nvlinkBwMode reference of nvlink bandwidth mode + * @return + * - \ref NVML_SUCCESS on success + * - \ref NVML_ERROR_INVALID_ARGUMENT if an invalid pointer is provided + * - \ref NVML_ERROR_NOT_SUPPORTED if GPU is not Hopper or newer architecture. + * - \ref NVML_ERROR_NO_PERMISSION if not root user + */ +nvmlReturn_t DECLDIR nvmlSystemGetNvlinkBwMode(unsigned int *nvlinkBwMode); + +/** + * Set new power limit of this device. + * + * For Kepler &tm; or newer fully supported devices. + * Requires root/admin permissions. + * + * See \ref nvmlDeviceGetPowerManagementLimitConstraints to check the allowed ranges of values. + * + * See \ref nvmlPowerValue_v2_t for more information on the struct. + * + * \note Limit is not persistent across reboots or driver unloads. + * Enable persistent mode to prevent driver from unloading when no application is using the device. + * + * This API replaces nvmlDeviceSetPowerManagementLimit. It can be used as a drop-in replacement for the older version. + * + * @param device The identifier of the target device + * @param powerValue Power management limit in milliwatts to set + * + * @return + * - \ref NVML_SUCCESS if \a limit has been set + * - \ref NVML_ERROR_UNINITIALIZED if the library has not been successfully initialized + * - \ref NVML_ERROR_INVALID_ARGUMENT if \a device is invalid or \a powerValue is NULL or contains invalid values * - \ref NVML_ERROR_NOT_SUPPORTED if the device does not support this feature * - \ref NVML_ERROR_GPU_IS_LOST if the target GPU has fallen off the bus or is otherwise inaccessible * - \ref NVML_ERROR_UNKNOWN on any unexpected error + * + * @see NVML_FI_DEV_POWER_AVERAGE + * @see NVML_FI_DEV_POWER_INSTANT + * @see NVML_FI_DEV_POWER_MIN_LIMIT + * @see NVML_FI_DEV_POWER_MAX_LIMIT + * @see NVML_FI_DEV_POWER_CURRENT_LIMIT */ -nvmlReturn_t DECLDIR nvmlDeviceGetProcessUtilization(nvmlDevice_t device, nvmlProcessUtilizationSample_t *utilization, - unsigned int *processSamplesCount, unsigned long long lastSeenTimeStamp); +nvmlReturn_t DECLDIR nvmlDeviceSetPowerManagementLimit_v2(nvmlDevice_t device, nvmlPowerValue_v2_t *powerValue); -/** @} */ /** * NVML API versioning support */ -#if defined(__NVML_API_VERSION_INTERNAL) + +#ifdef NVML_NO_UNVERSIONED_FUNC_DEFS +nvmlReturn_t DECLDIR nvmlInit(void); +nvmlReturn_t DECLDIR nvmlDeviceGetCount(unsigned int *deviceCount); +nvmlReturn_t DECLDIR nvmlDeviceGetHandleByIndex(unsigned int index, nvmlDevice_t *device); +nvmlReturn_t DECLDIR nvmlDeviceGetHandleByPciBusId(const char *pciBusId, nvmlDevice_t *device); +nvmlReturn_t DECLDIR nvmlDeviceGetPciInfo(nvmlDevice_t device, nvmlPciInfo_t *pci); +nvmlReturn_t DECLDIR nvmlDeviceGetPciInfo_v2(nvmlDevice_t device, nvmlPciInfo_t *pci); +nvmlReturn_t DECLDIR nvmlDeviceGetNvLinkRemotePciInfo(nvmlDevice_t device, unsigned int link, nvmlPciInfo_t *pci); +nvmlReturn_t DECLDIR nvmlDeviceGetGridLicensableFeatures(nvmlDevice_t device, nvmlGridLicensableFeatures_t *pGridLicensableFeatures); +nvmlReturn_t DECLDIR nvmlDeviceGetGridLicensableFeatures_v2(nvmlDevice_t device, nvmlGridLicensableFeatures_t *pGridLicensableFeatures); +nvmlReturn_t DECLDIR nvmlDeviceGetGridLicensableFeatures_v3(nvmlDevice_t device, nvmlGridLicensableFeatures_t *pGridLicensableFeatures); +nvmlReturn_t DECLDIR nvmlDeviceRemoveGpu(nvmlPciInfo_t *pciInfo); +nvmlReturn_t DECLDIR nvmlEventSetWait(nvmlEventSet_t set, nvmlEventData_t * data, unsigned int timeoutms); +nvmlReturn_t DECLDIR nvmlDeviceGetAttributes(nvmlDevice_t device, nvmlDeviceAttributes_t *attributes); +nvmlReturn_t DECLDIR nvmlComputeInstanceGetInfo(nvmlComputeInstance_t computeInstance, nvmlComputeInstanceInfo_t *info); +nvmlReturn_t DECLDIR nvmlDeviceGetComputeRunningProcesses(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_v1_t *infos); +nvmlReturn_t DECLDIR nvmlDeviceGetComputeRunningProcesses_v2(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_v2_t *infos); +nvmlReturn_t DECLDIR nvmlDeviceGetGraphicsRunningProcesses(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_v1_t *infos); +nvmlReturn_t DECLDIR nvmlDeviceGetGraphicsRunningProcesses_v2(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_v2_t *infos); +nvmlReturn_t DECLDIR nvmlDeviceGetMPSComputeRunningProcesses(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_v1_t *infos); +nvmlReturn_t DECLDIR nvmlDeviceGetMPSComputeRunningProcesses_v2(nvmlDevice_t device, unsigned int *infoCount, nvmlProcessInfo_v2_t *infos); +nvmlReturn_t DECLDIR nvmlDeviceGetGpuInstancePossiblePlacements(nvmlDevice_t device, unsigned int profileId, nvmlGpuInstancePlacement_t *placements, unsigned int *count); +nvmlReturn_t DECLDIR nvmlVgpuInstanceGetLicenseInfo(nvmlVgpuInstance_t vgpuInstance, nvmlVgpuLicenseInfo_t *licenseInfo); +#endif // #ifdef NVML_NO_UNVERSIONED_FUNC_DEFS + +#if defined(NVML_NO_UNVERSIONED_FUNC_DEFS) +// We don't define APIs to run new versions if this guard is present so there is +// no need to undef +#elif defined(__NVML_API_VERSION_INTERNAL) +#undef nvmlDeviceGetGraphicsRunningProcesses +#undef nvmlDeviceGetComputeRunningProcesses +#undef nvmlDeviceGetMPSComputeRunningProcesses +#undef nvmlDeviceGetAttributes +#undef nvmlComputeInstanceGetInfo +#undef nvmlEventSetWait +#undef nvmlDeviceGetGridLicensableFeatures +#undef nvmlDeviceRemoveGpu #undef nvmlDeviceGetNvLinkRemotePciInfo #undef nvmlDeviceGetPciInfo #undef nvmlDeviceGetCount #undef nvmlDeviceGetHandleByIndex #undef nvmlDeviceGetHandleByPciBusId #undef nvmlInit +#undef nvmlBlacklistDeviceInfo_t +#undef nvmlGetBlacklistDeviceCount +#undef nvmlGetBlacklistDeviceInfoByIndex +#undef nvmlDeviceGetGpuInstancePossiblePlacements +#undef nvmlVgpuInstanceGetLicenseInfo +#undef nvmlDeviceSetPowerManagementLimit + #endif #ifdef __cplusplus diff --git a/src/3rdparty/nvml/repo.json b/src/3rdparty/nvml/repo.json index 8612ffb39..ae0121463 100644 --- a/src/3rdparty/nvml/repo.json +++ b/src/3rdparty/nvml/repo.json @@ -1,6 +1,6 @@ { - "home": "https://developer.nvidia.com/gpu-deployment-kit", + "home": "https://github.com/NVIDIA/nvidia-settings/blob/main/src/nvml.h", "license": "Embed in source", - "version": "v9", + "version": "545.23.06", "author": "NVIDIA Corporation" } diff --git a/src/detection/gpu/gpu_linux.c b/src/detection/gpu/gpu_linux.c index fc891fc27..651cc0e42 100644 --- a/src/detection/gpu/gpu_linux.c +++ b/src/detection/gpu/gpu_linux.c @@ -243,7 +243,11 @@ static void pciHandleDevice(FF_MAYBE_UNUSED const FFGPUOptions* options, FFlist* #endif if (gpu->temperature != gpu->temperature && gpu->vendor.chars == FF_GPU_VENDOR_NAME_NVIDIA) - ffDetectNvidiaGpuTemp(&gpu->temperature, device->device_id); + { + char pciDeviceId[32]; + snprintf(pciDeviceId, sizeof(pciDeviceId) - 1, "%04x:%02x:%02x.%d", device->domain, device->bus, device->dev, device->func); + ffDetectNvidiaGpuTemp(&gpu->temperature, pciDeviceId, 0, 0); + } } jmp_buf pciInitJmpBuf; diff --git a/src/detection/gpu/gpu_windows.c b/src/detection/gpu/gpu_windows.c index 80ed6fd91..3ddba91fc 100644 --- a/src/detection/gpu/gpu_windows.c +++ b/src/detection/gpu/gpu_windows.c @@ -75,13 +75,6 @@ const char* ffDetectGPUImpl(FF_MAYBE_UNUSED const FFGPUOptions* options, FFlist* gpu->dedicated.total = dedicatedVideoMemory + dedicatedSystemMemory; gpu->shared.total = sharedSystemMemory; } - - if (options->temp && gpu->vendor.chars == FF_GPU_VENDOR_NAME_NVIDIA) - { - uint32_t deviceId; - if(ffRegReadUint(hDirectxKey, L"DeviceId", &deviceId, NULL)) - ffDetectNvidiaGpuTemp(&gpu->temperature, deviceId); - } } else if (!ffRegReadUint64(hKey, L"HardwareInformation.qwMemorySize", &gpu->dedicated.total, NULL)) { @@ -91,6 +84,15 @@ const char* ffDetectGPUImpl(FF_MAYBE_UNUSED const FFGPUOptions* options, FFlist* gpu->type = gpu->dedicated.total > 1024 * 1024 * 1024 ? FF_GPU_TYPE_DISCRETE : FF_GPU_TYPE_INTEGRATED; } } + + if (options->temp && gpu->vendor.chars == FF_GPU_VENDOR_NAME_NVIDIA) + { + uint32_t vendorId, deviceId, subSystemId; + // See: https://download.nvidia.com/XFree86/Linux-x86_64/545.23.06/README/supportedchips.html + // displayDevice.DeviceID = MatchingDeviceId "PCI\\VEN_10DE&DEV_2782&SUBSYS_513417AA&REV_A1" + if (swscanf(displayDevice.DeviceID, L"PCI\\VEN_%x&DEV_%x&SUBSYS_%x", &vendorId, &deviceId, &subSystemId) == 3) + ffDetectNvidiaGpuTemp(&gpu->temperature, NULL, (deviceId << 16) | vendorId, subSystemId); + } } return NULL; diff --git a/src/detection/temps/temps_nvidia.c b/src/detection/temps/temps_nvidia.c index 90f241d1a..61194f3a1 100644 --- a/src/detection/temps/temps_nvidia.c +++ b/src/detection/temps/temps_nvidia.c @@ -6,13 +6,17 @@ struct FFNvmlData { FF_LIBRARY_SYMBOL(nvmlDeviceGetCount_v2) FF_LIBRARY_SYMBOL(nvmlDeviceGetHandleByIndex_v2) + FF_LIBRARY_SYMBOL(nvmlDeviceGetHandleByPciBusId_v2) FF_LIBRARY_SYMBOL(nvmlDeviceGetPciInfo_v3) FF_LIBRARY_SYMBOL(nvmlDeviceGetTemperature) bool inited; } nvmlData; -const char* ffDetectNvidiaGpuTemp(double* temp, uint32_t deviceId) +// Use pciBusId if not NULL; use pciDeviceId and pciSubSystemId otherwise +// pciBusId = "domain:bus:device.function" +// pciDeviceId = (deviceId << 16) | vendorId +const char* ffDetectNvidiaGpuTemp(double* temp, const char* pciBusId, uint32_t pciDeviceId, uint32_t pciSubSystemId) { if (!nvmlData.inited) { @@ -29,6 +33,7 @@ const char* ffDetectNvidiaGpuTemp(double* temp, uint32_t deviceId) FF_LIBRARY_LOAD_SYMBOL_MESSAGE(libnvml, nvmlShutdown) FF_LIBRARY_LOAD_SYMBOL_VAR_MESSAGE(libnvml, nvmlData, nvmlDeviceGetCount_v2) FF_LIBRARY_LOAD_SYMBOL_VAR_MESSAGE(libnvml, nvmlData, nvmlDeviceGetHandleByIndex_v2) + FF_LIBRARY_LOAD_SYMBOL_VAR_MESSAGE(libnvml, nvmlData, nvmlDeviceGetHandleByPciBusId_v2) FF_LIBRARY_LOAD_SYMBOL_VAR_MESSAGE(libnvml, nvmlData, nvmlDeviceGetPciInfo_v3) FF_LIBRARY_LOAD_SYMBOL_VAR_MESSAGE(libnvml, nvmlData, nvmlDeviceGetTemperature) @@ -44,30 +49,40 @@ const char* ffDetectNvidiaGpuTemp(double* temp, uint32_t deviceId) if (nvmlData.ffnvmlDeviceGetTemperature == NULL) return "loading nvml library failed"; - uint32_t count; - if (nvmlData.ffnvmlDeviceGetCount_v2(&count) != NVML_SUCCESS) - return "nvmlDeviceGetCount_v2() failed"; - - for (uint32_t i = 0; i < count; i++) + nvmlDevice_t device = NULL; + if (pciBusId) { - nvmlDevice_t device; - if (nvmlData.ffnvmlDeviceGetHandleByIndex_v2(i, &device) != NVML_SUCCESS) - continue; + nvmlReturn_t ret = nvmlData.ffnvmlDeviceGetHandleByPciBusId_v2(pciBusId, &device); + if (ret != NVML_SUCCESS) + return "nvmlDeviceGetHandleByPciBusId_v2() failed"; + } + else + { + uint32_t count; + if (nvmlData.ffnvmlDeviceGetCount_v2(&count) != NVML_SUCCESS) + return "nvmlDeviceGetCount_v2() failed"; - nvmlPciInfo_t pciInfo; - if (nvmlData.ffnvmlDeviceGetPciInfo_v3(device, &pciInfo) != NVML_SUCCESS) - continue; + for (uint32_t i = 0; i < count; i++, device = NULL) + { + if (nvmlData.ffnvmlDeviceGetHandleByIndex_v2(i, &device) != NVML_SUCCESS) + continue; - if (pciInfo.pciDeviceId >> 16 != deviceId) - continue; + nvmlPciInfo_t pciInfo; + if (nvmlData.ffnvmlDeviceGetPciInfo_v3(device, &pciInfo) != NVML_SUCCESS) + continue; - uint32_t value; - if (nvmlData.ffnvmlDeviceGetTemperature(device, NVML_TEMPERATURE_GPU, &value) != NVML_SUCCESS) - return "nvmlDeviceGetTemperature(device, NVML_TEMPERATURE_GPU, &value) failed"; + if (pciInfo.pciDeviceId != pciDeviceId || pciInfo.pciSubSystemId != pciSubSystemId) + continue; - *temp = value; - return NULL; + break; + } + if (!device) return "Device not found"; } - return "Device not found"; + uint32_t value; + if (nvmlData.ffnvmlDeviceGetTemperature(device, NVML_TEMPERATURE_GPU, &value) != NVML_SUCCESS) + return "nvmlDeviceGetTemperature(device, NVML_TEMPERATURE_GPU, &value) failed"; + + *temp = value; + return NULL; } diff --git a/src/detection/temps/temps_nvidia.h b/src/detection/temps/temps_nvidia.h index 1b79d2522..3cf7dadda 100644 --- a/src/detection/temps/temps_nvidia.h +++ b/src/detection/temps/temps_nvidia.h @@ -2,4 +2,4 @@ #include -const char* ffDetectNvidiaGpuTemp(double* temp, uint32_t deviceId); +const char* ffDetectNvidiaGpuTemp(double* temp, const char* pciBusId, uint32_t pciDeviceId, uint32_t pciSubSystemId);