Change Log

This chapter list changes in API and bug fixes that were introduced to the library.

Changes between NVML v610 and v615

• Added NVML_GPU_RECOVERY_ACTION_BUS_RESET and NVML_GPU_RECOVERY_ACTION_SYSTEM_REBOOT to nvmlDeviceGpuRecoveryAction_t.

• Added GPU Operational Event APIs to register and wait for structured events and retrieve their associated context records:

  • nvmlEventSetRegisterGpuOperationalEvents_v1() and nvmlEventSetWait_v3()

  • nvmlEventSetGetContextCount_v1(), nvmlEventSetGetContextInfo_v1(), nvmlEventSetGetContextData_v1(), and nvmlEventSetGetGpuOperationalEventContextLegacyXid_v1()

• Added nvmlDeviceGetBankRemapperStatus_v1() to query the bank remapper status.

• Added a new temperature sensor type NVML_TEMPERATURE_GPU_MAX to report the temperature from the hottest part of the GPU.

• Added 2 new clock event reason bitmask values for board limit and reliability policies:

  • nvmlClocksEventReasonBoardLimit

  • nvmlClocksEventReasonReliability

• Added new NVML_NVLINK_STATE_ACTIVE_TRAFFIC_DISABLED state.

• Added nvmlDeviceSetNvlinkBwModeAsync_v1() to asynchronously set Nvlink Bandwidth mode.

• Added nvmlDeviceSetMemoryLimits_v1() to set device memory limits in a cgroup partition.

• Added nvmlDeviceGetMemoryLimits_v1() to get device memory limits in a cgroup partition.

• Modified nvmlDeviceGetMemoryInfo_v2() to account for cgroup memory partitions when called within the partition.

• Added nvmlDeviceSetAdaptiveTgpMode_v1() and nvmlDeviceGetAdaptiveTgpModeInfo_v1() to set and query Adaptive TGP mode.

• Added nvmlDeviceGetNvLinkTelemetrySamples_v1() to get historical NVLink telemetry samples.

• Modified nvmlDeviceWorkloadPowerProfileUpdateProfiles_v1() to allow non-privileged access using WPPS capability.

• Added new Workload Power Profile types to nvmlPowerProfileType_t.

• Added GPM metrics NVML_GPM_METRIC_NVLINK_L[36..71].

• Added missing comments on GPM metrics of C2C, cache, NVENC, GR CTXSW, and NVLink PER_SEC.

• Added new GPU fabric clique types and structures for per-type clique information:

  • NVML_GPU_FABRIC_CLIQUE_MAX

  • NVML_GPU_FABRIC_CLIQUE_TYPE_UNICAST_POINTER, NVML_GPU_FABRIC_CLIQUE_TYPE_MULTICAST_POINTER, NVML_GPU_FABRIC_CLIQUE_TYPE_UNICAST_LOGICAL_ENDPOINT, NVML_GPU_FABRIC_CLIQUE_TYPE_MULTICAST_LOGICAL_ENDPOINT

  • nvmlGpuFabricClique_v1_t

• Added nvmlDeviceGetGpuFabricInfo_v4() to retrieve per-type fabric clique assignments.

• Deprecated nvmlGpuFabricInfo_v3_t in favor of nvmlGpuFabricInfo_v4_t.

• Added new NVML_GPU_FABRIC_HEALTH_MASK_GFM_STATE* Health Mask flags to GPU Fabric Health Mask.

Changes between NVML v595 and v610

• Added new MCLK field IDs:

  • NVML_FI_DEV_MCLK_SWITCH_TYPE

  • NVML_FI_DEV_MCLK_MIN_SWITCH_INTERVAL_MILLISECONDS

• Added new NVLINK field IDs:

  • NVML_FI_DEV_NVLINK_PLR_XMIT_BLOCKS

  • NVML_FI_DEV_NVLINK_PLR_XMIT_RETRY_BLOCKS

• Added nvmlDeviceGetAccountingStats_v2() to retrieve additional utilization information:

  • sampleCount

  • sumGpuUtil

  • sumFbUtil

• Added a new NVLINK field ID:

  • NVML_FI_DEV_NVLINK_GET_DATA_RATE

• Added support for new power scope NVML_POWER_SCOPE_GPU_BASE, which represents GPU base power.

  • Added support in Power Management APIs:

    • nvmlDeviceSetPowerManagementLimit_v2()

    • nvmlDeviceGetPowerManagementLimit_v2

  • Added support for the following nvmlFieldValue_t fields with the new scope:

    • NVML_FI_DEV_POWER_MIN_LIMIT

    • NVML_FI_DEV_POWER_MAX_LIMIT

    • NVML_FI_DEV_POWER_DEFAULT_LIMIT

    • NVML_FI_DEV_POWER_CURRENT_LIMIT

    • NVML_FI_DEV_POWER_REQUESTED_LIMIT

• Added support for detecting nvmlClocksEventReasonSwThermalSlowdown on Thor Jetson platforms through:

  • nvmlDeviceGetSupportedClocksEventReasons()

  • nvmlDeviceGetCurrentClocksEventReasons()

• Added NVML_FI_DEV_MMA_STALL_PERCENT for MMA stall telemetry.

• Updated field ID NVML_FI_DEV_MMA_STALL_PERCENT value type from NVML_VALUE_TYPE_UNSIGNED_INT to NVML_VALUE_TYPE_DOUBLE.

• Added nvmlDevicePerfMetricsGetSamples_v1() to get Performance Metric samples.

• Added NVML_FI_PWR_SMOOTHING_SOC_POWER_SMOOTHING_ENABLED for State-Of-Charge Power Smoothing support.

• Added public API support for Common Platform Error Record (CPER) data:

  • nvmlSystemGetCPER_v1() to retrieve CPER records using a caller-supplied buffer and cursor-based iteration.

  • nvmlGetCPER_v1_t for input and output values, including cursor, buffer, and buffer size.

  • nvmlCPERCursor_v1_t for query filters and read position, including cperTypeMask, uuid, and handle.

  • nvmlCPERType_t and NVML_CPER_ACCESS_TYPE_GPU for record type filtering.

  • NVML_CPER_CURSOR_HANDLE_INIT for cursor initialization.

  • Added size-query support by calling with buffer set to NULL and bufferSize set to 0. The API returns NVML_ERROR_INSUFFICIENT_SIZE with the required size, or NVML_SUCCESS with bufferSize set to 0 when no records exist.

Changes between NVML v590 and v595

• Added new NVLINK field IDs:

  • NVML_FI_DEV_NVLINK_COUNT_RAW_ERRORS_LANE0

  • NVML_FI_DEV_NVLINK_COUNT_RAW_ERRORS_LANE1

  • NVML_FI_DEV_NVLINK_COUNT_RAW_BER_LANE0_V2

  • NVML_FI_DEV_NVLINK_COUNT_RAW_BER_LANE1_V2

  • NVML_FI_DEV_NVLINK_COUNT_RAW_BER_V2

• Updated the descriptions of nvmlGpmSampleGet() and nvmlGpmQueryDeviceSupport() to emphasize that both device and MIG handles are supported by these APIs.

• Added NVML_INIT_FLAG_FORCE_INIT to force GPU initialization when a previous nvmlInit was called with NO_GPUS and NO_ATTACH flags.

• Updated the GPM metrics’ name from NVLINK1_L[0..17] to NVLINK_L[18..35].

• Updated Nvlink Version to include NVML_NVLINK_VERSION_6_0.

• Added a new API to retrieve GPU PRM counters: nvmlDeviceReadPRMCounters_v1().

• Added new NVML_GPU_RECOVERY_ACTION_RECOVER_IMEX_DOMAIN enum to nvmlDeviceGpuRecoveryAction_s.

• Added nvmlDeviceGetBBXTimeData_v1(), which returns the cumulative number of seconds the GPU has had the driver loaded.

• Added new GPM UTIL metric NVML_GPM_METRIC_DMMA_TENSOR_UTIL.

• Added new GPM raw metrics.

• Added nvmlDeviceGetVgpuSchedulerState_v2() to fetch the vGPU software scheduler state without taking ARR mode into account.

• Added nvmlGpuInstanceGetVgpuSchedulerState_v2() to fetch the vGPU software scheduler state for the given GPU Instance without taking ARR mode into account.

• Added nvmlDeviceGetVgpuSchedulerLog_v2() to query new field weight in the vGPU software scheduler logs for the device.

• Added nvmlGpuInstanceGetVgpuSchedulerLog_v2() to query new field weight in the vGPU software scheduler logs for the given GPU Instance.

• Added nvmlDeviceSetVgpuSchedulerState_v2() to set the vGPU scheduler state, with support for disabling ARR mode removed in this version.

• Added nvmlGpuInstanceSetVgpuSchedulerState_v2() to set the vGPU scheduler state for the given GPU Instance, with support for disabling ARR mode removed in this version.

• Deprecated vGPU scheduler APIs:

  • nvmlDeviceGetVgpuSchedulerState()

  • nvmlGpuInstanceGetVgpuSchedulerState()

  • nvmlDeviceGetVgpuSchedulerLog()

  • nvmlGpuInstanceGetVgpuSchedulerLog()

  • nvmlDeviceSetVgpuSchedulerState()

  • nvmlGpuInstanceSetVgpuSchedulerState()

• Added nvmlDeviceVgpuForceGspUnload() to force GSP unload on a vGPU host.

• Added new Incorrect Configuration Status to nvmlGpuFabricInfoV_t:

  • NVML_GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_GPU_STATE_INVALID

• Added new NVML_GPU_FABRIC_HEALTH_MASK_PARTITION_ASSIGNED* Health Mask flags to GPU Fabric Health Mask.

• Added vGPU ‘NVIDIA RTX Virtual Game Dev’ as a licensed product for vGPU.

• Added new MIG profile NVML_GPU_INSTANCE_PROFILE_3_SLICE_GFX for 3-slice graphics and media support.

Changes between v580 and v590

The following new functionality is exposed on NVIDIA display drivers version 590 Production or later.

• Added new product architecture NVML_DEVICE_ARCH_NPU3.

• Deprecated NVML_BRAND_NVIDIA_DLA product brand.

• Added new NVML_BRAND_NVIDIA_NPU product brand.

• Added nvmlDeviceWorkloadPowerProfileUpdateProfiles_v1() to clear, set, or clobber existing workload power profiles from a single API.

• Deprecated workload power profile APIs, which will be removed in CUDA 15.0:

  • nvmlDeviceWorkloadPowerProfileSetRequestedProfiles()

  • nvmlDeviceWorkloadPowerProfileClearRequestedProfiles()

• Added NVML_FI_DEV_EDPP_MULTIPLIER field value to report the current EDPp multiplier as a percentage.

• Added nvmlDeviceGetUnrepairableMemoryFlag_v1() to report whether or not there is unrepairable memory.

• Updated nvmlDeviceGetVgpuSchedulerLog() and nvmlGpuInstanceGetVgpuSchedulerState() to access the vGPU scheduler logs for the NVENC0 and NVENC1 engines.

• Updated NVML_NVLINK_MAX_LINKS to 36 to accommodate for more NvLinks in Rubin.

• Added NVLINK1 GPM metrics for Rubin.

• Added 14 new Power Smoothing field IDs for Delayed Power Smoothing:

  • NVML_FI_PWR_SMOOTHING_PRIMARY_POWER_FLOOR

  • NVML_FI_PWR_SMOOTHING_SECONDARY_POWER_FLOOR

  • NVML_FI_PWR_SMOOTHING_MIN_PRIMARY_FLOOR_ACT_OFFSET

  • NVML_FI_PWR_SMOOTHING_MIN_PRIMARY_FLOOR_ACT_POINT

  • NVML_FI_PWR_SMOOTHING_WINDOW_MULTIPLIER

  • NVML_FI_PWR_SMOOTHING_DELAYED_PWR_SMOOTHING_SUPPORTED

  • NVML_FI_PWR_SMOOTHING_PROFILE_SECONDARY_POWER_FLOOR

  • NVML_FI_PWR_SMOOTHING_PROFILE_PRIMARY_FLOOR_ACT_WIN_MULT

  • NVML_FI_PWR_SMOOTHING_PROFILE_PRIMARY_FLOOR_TAR_WIN_MULT

  • NVML_FI_PWR_SMOOTHING_PROFILE_PRIMARY_FLOOR_ACT_OFFSET

  • NVML_FI_PWR_SMOOTHING_ADMIN_OVERRIDE_SECONDARY_POWER_FLOOR

  • NVML_FI_PWR_SMOOTHING_ADMIN_OVERRIDE_PRIMARY_FLOOR_ACT_WIN_MULT

  • NVML_FI_PWR_SMOOTHING_ADMIN_OVERRIDE_PRIMARY_FLOOR_TAR_WIN_MULT

  • NVML_FI_PWR_SMOOTHING_ADMIN_OVERRIDE_PRIMARY_FLOOR_ACT_OFFSET

• Added four new Power Smoothing profile parameter IDs to set new Delayed Power Smoothing profile fields:

  • NVML_POWER_SMOOTHING_PROFILE_PARAM_SECONDARY_POWER_FLOOR

  • NVML_POWER_SMOOTHING_PROFILE_PARAM_PRIMARY_FLOOR_ACT_WIN_MULT

  • NVML_POWER_SMOOTHING_PROFILE_PARAM_PRIMARY_FLOOR_TAR_WIN_MULT

  • NVML_POWER_SMOOTHING_PROFILE_PARAM_PRIMARY_FLOOR_ACT_OFFSET

• Added NVML_INIT_FLAG_FORCE_INIT to force GPU initialization when a previous nvmlInit was called with NO_GPUS and NO_ATTACH flags.

Changes between v575 and v580

The following new functionality is exposed on NVIDIA display drivers version 580 Production or later.

• Added nvmlDeviceGetRemappedRows_v2() to query the number of inactive remapped rows.

• Added NVML_FI_DEV_REMAPPED_ROWS_COR_INACTIVE and NVML_FI_DEV_REMAPPED_ROWS_UNC_INACTIVE for nvmlDeviceGetFieldValues() to query the number of inactive remapped rows.

• Fixed bug with NVML_FI_PWR_SMOOTHING_* field value numbering, which was different than the v570 values.

• Adjusted NVML_FI_DEV_CLOCKS_EVENT_REASON_* and NVML_FI_DEV_POWER_SYNC_BALANCING_* field value numbering to resolve overlap with NVML_FI_PWR_SMOOTHING_* field values.

• Added nvmlDeviceWorkloadPowerProfileUpdateProfiles_v1() to clear, set, or clobber existing profiles from a single API.

• Added nvmlDeviceSetRusdSettings_v1() to set RUSD (read only user shared data buffer) polling mask.

• Added nvmlDeviceGetSramUniqueUncorrectedEccErrorCounts() to get the counts of SRAM unique uncorrected ECC errors.

• Deprecated Applications Clocks APIs, which will be removed in CUDA 14.0:

  • nvmlDeviceSetApplicationsClocks()

  • nvmlDeviceGetApplicationsClock()

  • nvmlDeviceGetDefaultApplicationsClock()

  • nvmlDeviceResetApplicationsClocks()

• Deprecated nvmlDeviceGetViolationStatus(), which will be removed in CUDA 14.0.

• Added nvmlDeviceGetNvLinkInfo() to query device NVLINK info.

• Added nvmlDeviceGetPdi() to retrieve the device GPU PDI.

• Added Multi-GPU mode NVLINK Encryption NVML_CC_SYSTEM_MULTIGPU_NVLE.

• Added V2 struct to nvmlDeviceGetNvLinkInfo() to query NVLINK Firmware info.

• Added nvmlDeviceReadWritePRM_v1() to retrieve GPU PRM register contents.

• Added nvmlDeviceGetAddressingMode() to retrieve the addressing mode for the device.

• Added nvmlDeviceGetRepairStatus() to get ECC status info.

• Added DLA device support.

• Added nvmlDeviceGetGpuInstanceProfileInfoByIdV(), which allows for MIG GPU instance profile info to be queried with profileId instead of profile name.

• Updated nvmlGpuFabricInfoV_t to v3 to include a new Health Summary field and new Incorrect Configuration statuses.

  • nvmlGpuFabricInfo_v2_t is deprecated and will be removed in a future release.

  • New Incorrect Configuration Statuses:

    • NVML_GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_INCOMPATIBLE_GPU_FW

    • NVML_GPU_FABRIC_HEALTH_MASK_INCORRECT_CONFIGURATION_INVALID_LOCATION

• Added nvmlDeviceGetPowerMizerMode_v1() to query the current and supported power mizer modes on Maxwell and newer GPUs. Power mizer mode provides a hint to the driver as to how to manage GPU performance.

• Added nvmlDeviceSetPowerMizerMode_v1() to set the power mizer mode on Maxwell and newer GPUs.

• Added nvmlDeviceSetHostname_v1() and nvmlDeviceGetHostname_v1() to allow custom GPU hostname configuration.

Changes between v570 Update and v575

The following new functionality is exposed on NVIDIA display drivers version 575 Production or later.

• Added nvmlSystemEventSetCreate() to create a system event set.

• Added nvmlSystemEventSetFree() to release a system event set.

• Added nvmlSystemRegisterEvents() to register system events on a system event set.

• Added nvmlSystemEventSetWait() to wait for system event notification and obtain system event data.

• Added nvmlGpuInstanceGetCreatableVgpus() to query the currently creatable vGPU types on the user-provided GPU Instance.

• Added nvmlVgpuTypeGetMaxInstancesPerGpuInstance() to query the maximum number of vGPU instances per GPU Instance for the given vGPU type.

• Added nvmlGpuInstanceSetVgpuSchedulerState() to set the vGPU scheduler state for the given GPU Instance.

• Added nvmlGpuInstanceGetActiveVgpus() to query the currently active vGPU instances on the user-provided GPU Instance.

• Added nvmlGpuInstanceGetVgpuSchedulerState() to query the vGPU software scheduler state for the given GPU Instance.

• Added nvmlGpuInstanceGetVgpuSchedulerLog() to query the vGPU software scheduler logs for the given GPU Instance.

• Added nvmlGpuInstanceGetVgpuTypeCreatablePlacements() to query the creatable vGPU placement IDs of the vGPU type within a GPU Instance.

• Added nvmlGpuInstanceSetVgpuHeterogeneousMode() to enable or disable vGPU heterogeneous mode for the GPU Instance.

• Added nvmlGpuInstanceGetVgpuHeterogeneousMode() to query the vGPU heterogeneous mode for the GPU Instance.

• Updated nvmlDeviceGetVgpuCapabilities() to report whether GPU supports timesliced vGPU on MIG and whether MIG timesliced mode is enabled or not.

• Updated nvmlDeviceSetVgpuCapabilities() to set the MIG timesliced mode vGPU capability of a device.

• Updated nvmlDeviceSetVgpuHeterogeneousMode() to return NVML_ERROR_NOT_SUPPORTED when in MIG mode.

• Updated nvmlDeviceGetVgpuHeterogeneousMode() to return NVML_ERROR_NOT_SUPPORTED when in MIG mode.

• Updated nvmlDeviceGetVgpuTypeCreatablePlacements() to return NVML_ERROR_NOT_SUPPORTED when in MIG mode.

• Updated nvmlDeviceGetVgpuSchedulerLog() to return NVML_ERROR_NOT_SUPPORTED when in MIG mode.

• Updated nvmlDeviceSetVgpuSchedulerState() to return NVML_ERROR_NOT_SUPPORTED when in MIG mode.

• Updated nvmlDeviceGetVgpuSchedulerState() to return NVML_ERROR_NOT_SUPPORTED when in MIG mode.

• Added 3 new NVML_FI_DEV_C2C_LINK_ERROR field IDs:

  • NVML_FI_DEV_C2C_LINK_ERROR_INTR

  • NVML_FI_DEV_C2C_LINK_ERROR_REPLAY

  • NVML_FI_DEV_C2C_LINK_ERROR_REPLAY_B2B

• Added NVML_FI_DEV_C2C_LINK_POWER_STATE field ID.

• Added new CTXSW GPM Metrics.

• Added nvmlDeviceGetHandleByUUIDV() that supports both ASCII and binary format UUID to retrieve the device handle.

• Added 2 new NVML_FI_DEV_POWER_SYNC_BALANCING field IDs:

  • NVML_FI_DEV_POWER_SYNC_BALANCING_FREQ

  • NVML_FI_DEV_POWER_SYNC_BALANCING_AF

• Added 5 new Clock Event Reason Counters field IDs:

  • NVML_FI_DEV_CLOCKS_EVENT_REASON_SW_POWER_CAP

  • NVML_FI_DEV_CLOCKS_EVENT_REASON_SYNC_BOOST

  • NVML_FI_DEV_CLOCKS_EVENT_REASON_SW_THERM_SLOWDOWN

  • NVML_FI_DEV_CLOCKS_EVENT_REASON_HW_THERM_SLOWDOWN

  • NVML_FI_DEV_CLOCKS_EVENT_REASON_HW_POWER_BRAKE_SLOWDOWN

• Updated nvmlDeviceGetMemoryErrorCounter() to better account for transient vs. permanent errors.

• Added MIG profiles that can allocate all or none of Decoder, Encoder, JPEG, and OFA engines.

Changes between v565 and v570

The following new functionality is exposed on NVIDIA display drivers version 570 Production or later.

• Added field values for data related to Power Smoothing

• Added nvmlDevicePowerSmoothingActivatePresetProfile() to activate a specific Preset Profile for Power Smoothing

• Added nvmlDevicePowerSmoothingSetState() to enable/disable the Power Smoothing feature

• Added nvmlDevicePowerSmoothingUpdatePresetProfileParam() to update parameters to preset profiles for Power Smoothing

• Added new enums for fieldId NVML_FI_DEV_NVLINK_GET_STATE to expose INACTIVE, ACTIVE, and SLEEP state for a link

• Added nvmlDeviceGetMarginTemperature() to retrieve the thermal margin temperature (distance to nearest slowdown threshold).

• Added nvmlDeviceGetNvlinkSupportedBwModes() to get all supported Nvlink Bandwidth modes

• Added nvmlDeviceGetNvlinkBwMode() to get the current Nvlink Bandwidth mode

• Added nvmlDeviceSetNvlinkBwMode() to set the Nvlink Bandwidth mode

• Added MIG profiles with support for graphics.

• Added support for new recovery action - NVML_GPU_RECOVERY_ACTION_DRAIN_AND_RESET

• Deprecated nvml fieldIds NVML_FI_DEV_RESET_STATUS and NVML_FI_DEV_DRAIN_AND_RESET_STATUS. Usee NVML_FI_DEV_GET_GPU_RECOVERY_ACTION instead

• Added nvmlDeviceGetDramEncryptionMode() and nvmlDeviceSetDramEncryptionMode() to query and configure DRAM Encryption Mode

• Added 3 new flags to GPU Fabric Health Mask:

  • NVML_GPU_FABRIC_HEALTH_MASK_SHIFT_ROUTE_RECOVERY

  • NVML_GPU_FABRIC_HEALTH_MASK_SHIFT_ROUTE_UNHEALTHY

  • NVML_GPU_FABRIC_HEALTH_MASK_SHIFT_ACCESS_TIMEOUT_RECOVERY

• Added new counters for Nvlink5

• NVML_FI_DEV_NVLINK_COUNT_EFFECTIVE_ERRORS to get sum of the number of errors in each Nvlink packet

• NVML_FI_DEV_NVLINK_COUNT_EFFECTIVE_BER to get Effective BER for effective errors

• NVML_FI_DEV_NVLINK_COUNT_FEC_HISTORY_0 to 15 to get count of symbol errors that are corrected

• Swapped the values of field IDs NVML_FI_DEV_IS_MIG_MODE_INDEPENDENT_MIG_QUERY_CAPABLE and NVML_FI_DEV_NVLINK_GET_POWER_THRESHOLD_MAX to fix backwards compatibility with v550.

• New revision of nvmlPlatformInfo_t -- nvmlPlatformInfo_v2 has been added. In this version the following fields from v1 have been renamed

  • rackGuid to chassisSerialNumber

  • chassisPhysicalSlotNumber to slotNumber

  • computeSlotIndex to trayIndex

  • nodeIndex to hostId

• nvmlPlatformInfo_v1 is deprecated and will be removed in subsequent releases

• Revert the fix for the issue where PCIe throughput (reported via nvmlDeviceGetPcieThroughput() and nvidia-smi -q) is 1000 times bigger than its actual value

Changes between v560 and v565

The following new functionality is exposed on NVIDIA display drivers version 565 Production or later.

• Fixed the ECC error count mismatch between nvidia-smi query output and NVML APIs, nvmlDeviceGetMemoryErrorCounter() and nvmlDeviceGetFieldValues().

• Added new value NVML_CC_SYSTEM_CPU_CAPS_AMD_SNP_VTOM for CC CPU capability reporting.

• Added nvmlDeviceGetCoolerInfo() to retrieve a cooler’s control signal characteristics and target that cooler cools.

• Added new value NVML_CC_SYSTEM_CPU_CAPS_AMD_SEV_SNP for CC CPU capability reporting.

• Added nvmlDeviceGetFanSpeedRPM() to report the intended operating speed in rotations per minute (RPM) of the device’s specified fan.

• Added nvmlDeviceGetPerformanceModes() to retrieve a performance modes string with all the performance modes defined for this device along with their associated GPU Clock and Memory Clock values.

• Added nvmlDeviceGetCurrentClockFreqs() to retrieve a string with the associated GPU Clock and Memory Clock values for the current pstate.

• Added nvmlNvlinkVersion_t enum to define NvLink Version.

• Added nvmlDeviceGetPlatformInfo() to retrieve the platform information of a device.

• Added new event type nvmlEventTypeGpuUnavailableError.

• Removed support for nvmlDeviceGetNvLinkCrcLaneErrorCounter, nvmlDeviceGetNvLinkEccLaneErrorCounter, and nvmlDeviceGetNvLinkErrorCounter() on Blackwell.

• Removed support for fieldIds NVML_FI_DEV_NVLINK_ERROR_DL_REPLAY, NVML_FI_DEV_NVLINK_ERROR_DL_RECOVERY, and NVML_FI_DEV_NVLINK_ERROR_DL_CRC on Blackwell.

• Added nvmlVgpuInstanceGetRuntimeStateSize() to get the vGPU runtime state size.

• Updated nvmlDeviceGetVgpuTypeSupportedPlacements() function to report both Heterogeneous and Homogeneous vGPU placements.

• Updated nvmlDeviceGetVgpuCapabilities to report the Homogeneous vGPU capability.

• Added new event type nvmlEventTypeGpuRecoveryAction.

• Added new fieldId to query GPU recovery action NVML_FI_DEV_GET_GPU_RECOVERY_ACTION.

• Deprecated fieldIds:

  • NVML_FI_DEV_NVLINK_COUNT_VL15_DROPPED to get Number of VL15 MADs dropped on a link in NVLink5

  • NVML_FI_DEV_NVLINK_COUNT_RAW_BER_LANE0 to get BER per lane for lane 0

  • NVML_FI_DEV_NVLINK_COUNT_RAW_BER_LANE1 to get BER per lane for lane 1

  • NVML_FI_DEV_NVLINK_COUNT_RAW_BER to get BER per link. Sum of all the raw errors per lane/Bits received per link

  • NVML_FI_DEV_NVLINK_COUNT_EFFECTIVE_ERRORS to get Sum of the number of errors in each Nvlink packet

  • NVML_FI_DEV_NVLINK_COUNT_EFFECTIVE_BER to get Effective BER for effective errors

Changes between v555 and v560

The following new functionality is exposed on NVIDIA display drivers version 560 Production or later.

• Added field values NVML_FI_DEV_PCIE_OUTBOUND_ATOMICS_MASK and NVML_FI_DEV_PCIE_INBOUND_ATOMICS_MASK for nvmlDeviceGetFieldValues().

• Added field IDs NVML_FI_DEV_RESET_STATUS and NVML_FI_DEV_DRAIN_AND_RESET_STATUS which correspond to the nvidia-smi output.

• Added NVML_DEVICE_ARCH_T23X architecture type.

• Added nvmlVgpuTypeGetBAR1Info() to query the BAR1 information of a vGPU type.

• Added new event types, nvmlEventTypeSingleBitEccErrorStorm, nvmlEventTypeDramRetirementEvent, nvmlEventTypeDramRetirementFailure, nvmlEventTypeNonFatalPoisonError and nvmlEventTypeFatalPoisonError.

• Added nvmlSystemGetDriverBranch() to query the driver branch information.

Changes between v550 and v555

The following new functionality is exposed on NVIDIA display drivers version 555 Production or later.

• Added nvmlDeviceGetClockOffsets() to query min, max and current clock offset value on a Maxwell and later GPU for a specified clock. Note: nvmlDeviceGetGpcClkVfOffset(), nvmlDeviceGetMemClkVfOffset(), nvmlDeviceGetGpcClkMinMaxVfOffset() and nvmlDeviceGetMemClkMinMaxVfOffset() will be deprecated in a future release. Use nvmlDeviceGetClockOffsets() instead.

• Added nvmlDeviceSetClockOffsets() to control clock offset value on a Maxwell and later GPU for a specified clock. Note: nvmlDeviceSetGpcClkVfOffset() and nvmlDeviceSetMemClkVfOffset() will be deprecated in a future release. Use nvmlDeviceSetClockOffsets() instead.

• Added two new field IDs NVML_FI_DEV_PCIE_COUNT_TX_BYTES and NVML_FI_DEV_PCIE_COUNT_RX_BYTES for nvmlDeviceGetFieldValues().

• Added new API nvmlDeviceGetCapabilities with the first capability bit NVML_DEV_CAP_EGM for Extended GPU Memory (EGM) capability.

• Added multiGpuMode display on CC enabled system via new API nvmlSystemGetConfComputeSettings() or “nvidia-smi conf-compute --get-multigpu-mode” or “nvidia-smi conf-compute -mgm”.

• Added new field ID NVML_FI_DEV_NVLINK_GET_POWER_THRESHOLD_MAX to get the Max Nvlink Power Threshold for a device.

Changes between v545 and v550

The following new functionality is exposed on NVIDIA display drivers version 550 Production or later.

• Added nvmlDeviceGetNumaNodeId() to query the NUMA node of a GPU.

• Added new GPM metric ID NVML_GPM_METRIC_NVOFA_1_UTIL to nvmlGpmMetricId_t.

• Added new field ID NVML_FI_DEV_IS_MIG_MODE_INDEPENDENT_MIG_QUERY_CAPABLE, to check MIG query capable device irrespective of MIG mode.

• Deprecated NVML_P2P_CAPS_INDEX_PROP and added NVML_P2P_CAPS_INDEX_PCI to reflect the same P2P capability.

• Added nvmlDeviceGetProcessesUtilizationInfo() to retrieve the recent utilization and process ID for all running processes.

• Added new struct nvmlProcessesUtilizationInfo_v1_t, which includes the new utilization of NVJPG and NVOFA.

• Added nvmlDeviceGetVgpuInstancesUtilizationInfo() to retrieve the recent utilization for vGPU instances running on a physical GPU.

• Added nvmlDeviceGetVgpuProcessesUtilizationInfo() to retrieve the recent utilization for processes running on vGPU instances on a physical GPU.

• Added nvmlDeviceSetVgpuHeterogeneousMode() to enable or disable vGPU heterogenous mode for the device.

• Added nvmlDeviceGetVgpuHeterogeneousMode() to query the vGPU heterogenous mode for the device.

• Added nvmlVgpuInstanceGetPlacementId() to query placement ID of the active vGPU instance.

• Added nvmlDeviceGetVgpuTypeSupportedPlacements() to query the supported vGPU placement IDs of a vGPU type.

• Added nvmlDeviceGetVgpuTypeCreatablePlacements() to query the creatable vGPU placement IDs of a vGPU type.

• Added support to display confidential compute protected memory along with fb and bar1 in nvidia-smi pmon and dmon commands.

• Added nvmlDeviceGetGpuFabricInfoV() to query GPU Fabric Probe Info for the device.

• Deprecated nvmlDeviceGetGpuFabricInfo(). This function should not be used, and will be removed in a future release. Use nvmlDeviceGetGpuFabricInfoV() instead.

• Modified nvmlDeviceGetGpuInstanceProfileInfo() and nvmlDeviceGetGpuInstancePossiblePlacements_v2() to no longer require MIG being enabled.

• Added new encoder type NVML_ENCODER_QUERY_AV1 and NVML_ENCODER_QUERY_UNKNOWN to enumeration nvmlEncoderType_t.

• Added nvmlSystemSetConfComputeKeyRotationThresholdInfo() to set confidential compute key rotation threshold.

• Added nvmlSystemGetConfComputeKeyRotationThresholdInfo() to query confidential compute key rotation threshold detail.

• Added nvmlDeviceSetVgpuCapabilities() to set the desirable vGPU capability of a device.

Changes between v535 and v545

The following new functionality is exposed on NVIDIA display drivers version 545 Production or later.

• Added a new error code NVML_ERROR_GPU_NOT_FOUND to be returned if no supported GPUS are found during initialization.

• In nvmlGpuFabricInfo_v2_t, partitionId has been renamed to cliqueId.

• Added new versioned structs nvmlGpuInstanceProfileInfo_v3_t and nvmlComputeInstanceProfileInfo_v3_t.

• Added nvmlDeviceGetLastBBXFlushTime() for retrieving the timestamp and duration of the latest flush of the BBX object to the inforom storage.

• Added NVML_POWER_SCOPE_MEMORY to report out power usage for GPU Memory.

• Added nvmlDeviceGetPciInfo_v3() which expands nvmlDeviceGetPciInfo to also report PCI base and sub classcodes.

• Added new struct nvmlPciInfoExt_v1_t, which is used in nvmlDeviceGetPciInfoExt().

• Added nvmlDeviceGetRunningProcessDetailList() API to get information about Compute, Graphics or MPS-Compute processes running on a GPU with protected memory usage info.

Changes between v530 and v535

The following new functionality is exposed on NVIDIA display drivers version 535 Production or later.

• Added nvmlDeviceGetSramEccErrorStatus() to query SRAM ECC error status for the device.

• Added nvmlDeviceGetModuleId() for getting device module ID.

• Updated nvmlDeviceGetPowerSource() API to report undersized power source.

• Added:cpp:func:nvmlDeviceGetJpgUtilization and nvmlDeviceGetOfaUtilization() APIs.

• Added nvmlSystemGetNvlinkBwMode() and nvmlSystemSetNvlinkBwMode() APIs.

• Added nvmlDeviceSetVgpuSchedulerState() to set the vGPU scheduler state.

• Added new field ID NVML_FI_DEV_IS_RESETLESS_MIG_SUPPORTED for device’s resetless MIG capability.

• Added nvmlDeviceGetComputeRunningProcesses_v3() to get information about Compute processes running on a GPU.

• Added nvmlDeviceGetGraphicsRunningProcesses_v3() to get information about Graphics processes running on a GPU.

• Added nvmlDeviceGetMPSComputeRunningProcesses_v3() to get information about MPS-Compute processes running on a GPU.

• Added nvmlDeviceGetRunningProcessDetailList() to get information about Compute, Graphics or MPS-Compute processes running on a GPU with protected memory usage info.

• Added nvmlDeviceGetLastBBXFlushTime() for retrieving the timestamp and duration of the latest flush of the BBX object to the inforom storage.

• Added new field ID NVML_FI_DEV_PCIE_COUNT_CORRECTABLE_ERRORS for PCIe correctable errors counter.

• Added new field ID NVML_FI_DEV_PCIE_COUNT_NAKS_RECEIVED for PCIe NAK Receive counter.

• Added new field ID NVML_FI_DEV_PCIE_COUNT_RECEIVER_ERROR for PCIe receiver error counter.

• Added new field ID NVML_FI_DEV_PCIE_COUNT_BAD_TLP for PCIe bad TLP counter.

• Added new field ID NVML_FI_DEV_PCIE_COUNT_NAKS_SENT for NAK Send counter.

• Added new field ID NVML_FI_DEV_PCIE_COUNT_BAD_DLLP for PCIe bad DLLP counter.

• Added new field ID NVML_FI_DEV_PCIE_COUNT_NON_FATAL_ERROR for PCIe non fatal error counter.

• Added new field ID NVML_FI_DEV_PCIE_COUNT_FATAL_ERROR for PCIe fatal error counter.

• Added new field ID NVML_FI_DEV_PCIE_COUNT_UNSUPPORTED_REQ for PCIe unsupported request counter.

• Added new field ID NVML_FI_DEV_PCIE_COUNT_LCRC_ERROR for PCIe LCRC error counter.

• Added new field ID NVML_FI_DEV_PCIE_COUNT_LANE_ERROR for per lane error counter with scope as PCIe lane number.

• Added nvmlDeviceGetPowerUsage_v2 to retrieve current power usage.

• Added nvmlDeviceGetTotalEnergyConsumption_v2 to get current energy consumption.

• Added nvmlDeviceSetPowerManagementLimit_v2() to set the power limit.

• Added new field IDs, NVML_FI_GPU_POWER_AVERAGE and NVML_FI_GPU_POWER_INSTANT, to query power usage.

• Renamed nvmlDeviceCcuGetStreamState to nvmlGpmQueryIfStreamingEnabled() and nvmlDeviceCcuSetStreamState to nvmlGpmSetStreamingEnabled().

• Added support to display confidential compute protected memory along with fb and bar1 in nvidia-smi pmon and dmon commands.

• Added new field IDs NVML_FI_DEV_TEMPERATURE_SHUTDOWN_TLIMIT, NVML_FI_DEV_TEMPERATURE_SLOWDOWN_TLIMIT, NVML_FI_DEV_TEMPERATURE_MEM_MAX_TLIMIT, and NVML_FI_DEV_TEMPERATURE_GPU_MAX_TLIMIT to query temperature thresholds on Ada and later architectures.

• Introduced ClockEventReasons and related APIs which should be used instead of ClockThrottleReasons. Deprecated ClockThrottleReasons.

• Added ability to get GPS Temperature Threshold with nvmlDeviceGetTemperatureThreshold() using the new enum NVML_TEMPERATURE_THRESHOLD_GPS_CURR.

Changes between v525 and v530

The following new functionality is exposed on NVIDIA display drivers version 530 Production or later.

• Fixed a typo in nvmlGpuP2PStatus_t: added a new enum entry for NVML_P2P_STATUS_CHIPSET_NOT_SUPPORTED with the same numeric value as the existing erroneous entry (”NVML_P2P_STATUS_CHIPSET_NOT_SUPPORED”).

• Added nvmlDeviceGetVgpuSchedulerLog() to fetch the vGPU software scheduler logs.

• Added nvmlDeviceGetVgpuSchedulerState() to fetch the vGPU software scheduler state.

• Added nvmlDeviceGetVgpuSchedulerCapabilities() to fetch the vGPU software scheduler capabilities.

Changes between v520 and v525

The following new functionality is exposed on NVIDIA display drivers version 525 Production or later.

• Added nvmlDeviceGetPcieAtomicCaps to report PCIe atomic capabilities.

• Added nvmlDeviceCcuGetStreamState API to report the counter collection unit stream state.

• Added nvmlDeviceCcuSetStreamState API to set the counter collection unit stream state.

• Removed support for NVML_FI_DEV_LINK_SPEED_MBPS_L{0..} field IDs in Hopper. Replaced with NVML_FI_DEV_NVLINK_GET_SPEED with scope as link ID.

• Removed support for NVML_FI_DEV_NVLINK_CRC_FLIT_ERROR_COUNT{0..} field IDs in Hopper. Replaced with NVML_FI_DEV_NVLINK_ERROR_DL_CRC with scope as link ID.

• Removed support for NVML_FI_DEV_NVLINK_REPLAY_ERROR_COUNT_L{0..} field IDs in Hopper. Replaced with NVML_FI_DEV_NVLINK_ERROR_DL_REPLAY with scope as link ID.

• Removed support for NVML_FI_DEV_NVLINK_RECOVERY_ERROR_COUNT_{0..} field IDs in Hopper. Replaced with NVML_FI_DEV_NVLINK_ERROR_DL_RECOVERY with scope as link ID.

• Added new field ID NVML_FI_DEV_NVLINK_GET_STATE to get nvlink state.

• Added new field ID NVML_FI_DEV_NVLINK_GET_VERSION to get nvlink version.

• Added new field ID NVML_FI_DEV_C2C_LINK_COUNT to get C2C link count.

• Added new field ID NVML_FI_DEV_C2C_LINK_GET_STATUS to get C2C link status.

• Added new field ID NVML_FI_DEV_C2C_LINK_GET_MAX_BW to get C2C link bandwidth.

Changes between v515 and v520

The following new functionality is exposed on NVIDIA display drivers version 520 Production or later.

• Added nvmlDeviceGetMemClkVfOffset() API to report the MemClk VF offset value.

• Added nvmlDeviceSetMemClkVfOffset() API to set the MemClk VF offset value.

• Added nvmlDeviceGetMemClkMinMaxVfOffset() API to report the Memory clock min and max VF offset that user can set for a specified GPU.

• Added nvmlDeviceGetTargetFanSpeed() API to report the intended target speed of the device’s specified fan.

• Added nvmlDeviceGetGpcClkMinMaxVfOffset() API to report the Graphics clock min and max VF offset that user can set for a specified GPU.

• Added nvmlGpmMetricsGet() to calculate GPM metrics from two GPM samples.

• Added nvmlGpmSampleFree() to free allocated GPM sample.

• Added nvmlGpmSampleAlloc() to allocate a GPM sample.

• Added nvmlGpmSampleGet() to retrieve a GPM snapshot.

• Added nvmlGpmQueryDeviceSupport() to query whether a device supports GPM

• Added nvmlDeviceGetSupportedPowerModes API to report the GPU’s supported power mode mask.

• Added nvmlDeviceGetPowerMode API to report the GPU’s current power mode.

• Added nvmlDeviceSetPowerMode API to set the new power mode.

• Added nvmlDeviceGetFanControlPolicy_v2() API to report the control policy for a specified GPU fan.

• Added nvmlDeviceSetFanControlPolicy() API to set the control policy for a specified GPU fan.

Changes between v510 and v515

The following new functionality is exposed on NVIDIA display drivers version 515 Production or later.

• Added nvmlDeviceGetDefaultECCMode API to report the GPU’s default ECC Mode.

• Added nvmlDeviceGetPcieSpeed() API to report the GPU’s PCIe link speed.

• Added nvmlDeviceGetDynamicPstatesInfo() API to report the GPU’s P-states information.

• Added nvmlDeviceSetFanSpeed_v2() API to set the GPU’s fan speed.

• Added nvmlDeviceSetDefaultFanSpeed_v2() API to set the GPU’s default fan speed.

• Added nvmlDeviceGetThermalSettings() API to report the GPU’s thermal system information.

• Added nvmlDeviceGetMinMaxClockOfPState() API to report the min and max clocks of some clock domain for a given PState.

• Added nvmlDeviceGetSupportedPerformanceStates() API to get all supported Performance States (P-States) for the GPU.

• Added nvmlDeviceGetGpcClkVfOffset() API to report the GPCCLK VF offset value.

• Added nvmlDeviceSetGpcClkVfOffset() API to set the GPCCLK VF offset value.

• Added nvmlDeviceGetMinMaxFanSpeed() API to report the min and max fan speed that user can set for a specified GPU fan.

Changes between v495 and v510

The following new functionality is exposed on NVIDIA display drivers version 510 Production or later.

• Added nvmlDeviceGetGpuInstanceProfileInfoV() and nvmlGpuInstanceGetComputeInstanceProfileInfoV() APIs to include the profile name in their output.

• Added nvmlDeviceGetMemoryBusWidth() API to report the GPU’s Memory Bus Width.

• Added nvmlDeviceGetPcieLinkMaxSpeed() API to report the GPU’s PCIe Max Speed.

• Added nvmlDeviceGetPowerSource() API to report the GPU’s power source as AC or battery.

• Added nvmlDeviceGetNumFans() API to report the GPU’s number of fans.

• Added nvmlDeviceGetNumGpuCores() API to report the GPU’s number of cores.

• Added nvmlDeviceGetMemoryInfo_v2(). The new version accounts separately for system-reserved memory, and includes it in the used memory amount. The previous version of the API reduced the total memory amount by the amount of system-reserved memory.

• Added nvmlDeviceGetAdaptiveClockInfoStatus() API to report the status of adaptive clocking for the GPU.

Changes between v465 and v470

The following new functionality is exposed on NVIDIA display drivers version 470 Production or later.

• Added new MIG GPU instance profile NVML_GPU_INSTANCE_PROFILE_1_SLICE_REV1.

• Added nvmlDeviceGetGpuInstancePossiblePlacements_v2(). The previous version of the API will not support the profiles with possible placements greater than its total capacity, such as NVML_GPU_INSTANCE_PROFILE_1_SLICE_REV1.

Changes between v460 and v465

The following new functionality is exposed on NVIDIA display drivers version 465 Production or later.

• Added new NVML_BRAND_* enumeration values for NVIDIA, NVIDIA_RTX, GEFORCE_RTX, QUADRO_RTX and TITAN_RTX.

• Updated nvmlDeviceGetHandleByUUID() to make it MIG-aware.

• Updated nvmlDeviceGetUUID() to return MIG UUIDs in the canonical format, ‘MIG-UUID’.

• Updated nvmlDeviceGetHandleByUUID() to accept both UUID formats, ‘MIG-UUID’ and ‘MIG-GPU UUID/GID/CID’.

• The nvmlDeviceSetAPIRestriction() and nvmlDeviceGetAPIRestriction() APIs would no longer support the ability to toggle root-only requirement for nvmlDeviceSetApplicationsClocks() and nvmlDeviceResetApplicationsClocks().

Changes between v450 and v460

The following new functionality is exposed on NVIDIA display drivers version 460 Production or later.

• Added nvmlDeviceCreateGpuInstanceWithPlacement() to allow placement specification when creating a new MIG GPU instance.

Changes between v445 and v450

The following new functionality is exposed on NVIDIA display drivers version 450 Production or later.

• Updated nvmlDeviceGetFanSpeed() and nvmlDeviceGetFanSpeed_v2() for allowing fan speeds greater than 100% to be reported.

• Added nvmlDeviceGetCpuAffinityWithinScope() to determine the closest processor(s) within a NUMA node or socket.

• Added nvmlDeviceGetMemoryAffinity() to determine the closest NUMA node(s) within a NUMA node or socket.

• Added support to query and disable MIG mode on Windows.

Changes between v418 and v445

The following new functionality is exposed on NVIDIA display drivers version 445 Production or later.

• Added support for the NVIDIA Ampere architecture.

• Added support for Multi Instance GPU management. Refer to the “Multi Instance GPU Management” section for details.

Changes between v361 and v418

The following new functionality is exposed on NVIDIA display drivers version 418 Production or later.

• Added support for the Volta and Turing architectures, bug fixes, performance improvements, and new features.

Changes between v349 and v361

The following new functionality is exposed on NVIDIA display drivers version 361 Production or later.

• Added nvmlDeviceGetBoardPartNumber() to return GPU part numbers

• Removed support for exclusive thread compute mode (Deprecated in 7.5)

• Added NVML_CLOCK_VIDEO (encoder/decoder) clock type as a supported clock type for nvmlDeviceGetClockInfo() and nvmlDeviceGetMaxClockInfo().

Changes between v346 and v349

The following new functionality is exposed on NVIDIA display drivers version 349 Production or later.

• Added nvmlDeviceGetTopologyCommonAncestor() to find the common path between two devices.

• Added nvmlDeviceGetTopologyNearestGpus() to get a set of GPUs given a path level.

• Added nvmlSystemGetTopologyGpuSet() to retrieve a set of GPUs with a given CPU affinity.

• Discontinued Perl bindings support.

• Updated nvmlDeviceGetAccountingPids(), nvmlDeviceGetAccountingBufferSize() and nvmlDeviceGetAccountingStats() to report accounting information for both active and terminated processes. The execution time field in nvmlAccountingStats_t structure is populated only when the process is terminated.

Changes between v340 and v346

The following new functionality is exposed on NVIDIA display drivers version 346 Production or later.

• Added nvmlDeviceGetGraphicsRunningProcesses_v2 to get information about Graphics Processes running on a device.

• Added nvmlDeviceGetPcieReplayCounter() to get PCI replay counters.

• Added nvmlDeviceGetPcieThroughput() to get PCI utilization information.

• Discontinued Perl bindings support.

Changes between NVML v331 and v340

The following new functionality is exposed on NVIDIA display drivers version 340 Production or later.

• Added nvmlDeviceGetSamples() to get recent power, utilization and clock samples for the GPU.

• Added nvmlDeviceGetTemperatureThreshold() to get temperature thresholds for the GPU.

• Added nvmlDeviceGetBrand() to get the brand name of the GPU.

• Added nvmlDeviceGetViolationStatus() to get the duration of time during which the device was throttled (lower than requested clocks) due to power or thermal constraints. Violations due to thermal capping is not supported at this time.

• Added nvmlDeviceGetEncoderUtilization() to get the GPU video encoder utilization.

• Added nvmlDeviceGetDecoderUtilization() to get the GPU video decoder utilization.

• Added nvmlDeviceGetCpuAffinity() to get the closest processor(s) affinity to a particular GPU.

• Added nvmlDeviceSetCpuAffinity() to set the affinity of a particular GPU to the closest processor.

• Added nvmlDeviceClearCpuAffinity() to clear the affinity of a particular GPU.

• Added nvmlDeviceGetBoardId() to get a unique boardId for the running system.

• Added nvmlDeviceGetMultiGpuBoard() to get whether the device is on a multiGPU board.

• Added nvmlDeviceGetAutoBoostedClocksEnabled() and nvmlDeviceSetAutoBoostedClocksEnabled() for querying and setting the state of auto boosted clocks on supporting hardware.

• Added nvmlDeviceSetDefaultAutoBoostedClocksEnabled() for setting the default state of auto boosted clocks on supporting hardware.

Changes between NVML v5.319 Update and v331

The following new functionality is exposed on NVIDIA display drivers version 331 or later.

• Added nvmlDeviceGetMinorNumber() to get the minor number for the device.

• Added nvmlDeviceGetBAR1MemoryInfo() to get BAR1 total, available and used memory size.

• Added nvmlDeviceGetBridgeChipInfo() to get the information related to bridge chip firmware.

• Added enforced power limit query API nvmlDeviceGetEnforcedPowerLimit()

• Updated nvmlEventSetWait to return xid event data in case of xid error event.

Changes between NVML v5.319 RC and v5.319 Update

The following new functionality is exposed on NVIDIA display drivers version 319 Update or later.

• Added nvmlDeviceSetAPIRestriction() and nvmlDeviceGetAPIRestriction(), with initial ability to toggle root-only requirement for nvmlDeviceSetApplicationsClocks() and nvmlDeviceResetApplicationsClocks().

Changes between NVML v4.304 Production and v5.319 RC

The following new functionality is exposed on NVIDIA display drivers version 319 RC or later.

• Added _v2 versions of nvmlDeviceGetHandleByIndex and nvmlDeviceGetCount that also count devices not accessible by current user

  • nvmlDeviceGetHandleByIndex_v2() (default) can also return NVML_ERROR_NO_PERMISSION

• Added nvmlInit_v2() and nvmlDeviceGetHandleByIndex_v2() that is safer and thus recommended function for initializing the library

  • nvmlInit_v2() lazily initializes only requested devices (queried with nvmlDeviceGetHandle*)

  • nvml.h defines nvmlInit_v2() and nvmlDeviceGetHandleByIndex_v2() as default functions

• Added nvmlDeviceGetIndex()

• Added NVML_ERROR_GPU_IS_LOST to report GPUs that have fallen off the bus.

  • All NVML device APIs can return this error code, as a GPU can fall off the bus at any time.

• Added new class of APIs for gathering process statistics (nvmlAccountingStats)

• Application Clocks are no longer supported on GPU’s from Quadro product line

• Added APIs to support dynamic page retirement. See nvmlDeviceGetRetiredPages() and nvmlDeviceGetRetiredPagesPendingStatus()

• Renamed nvmlClocksThrottleReasonUserDefinedClocks to nvmlClocksThrottleReasonApplicationsClocksSetting. Old name is deprecated and can be removed in one of the next major releases.

• Added nvmlDeviceGetDisplayActive() and updated documentation to clarify how it differs from nvmlDeviceGetDisplayMode()

Changes between NVML v4.304 RC and v4.304 Production

The following new functionality is exposed on NVIDIA display drivers version 304 Production or later.

• Added nvmlDeviceGetGpuOperationMode() and nvmlDeviceSetGpuOperationMode().

Changes between NVML v3.295 and v4.304 RC

The following new functionality is exposed on NVIDIA display drivers version 304 RC or later.

• Added nvmlDeviceGetInforomConfigurationChecksum() and nvmlDeviceValidateInforom().

• Added nvmlDeviceGetDisplayActive() and updated documentation to clarify how it differs from nvmlDeviceGetDisplayMode().

• Added new error return value for initialization failure due to kernel module not receiving interrupts.

• Added nvmlDeviceSetApplicationsClocks(), nvmlDeviceGetApplicationsClock(), nvmlDeviceResetApplicationsClocks().

• Added nvmlDeviceGetSupportedMemoryClocks() and nvmlDeviceGetSupportedGraphicsClocks().

• Added nvmlDeviceGetPowerManagementLimitConstraints(), nvmlDeviceGetPowerManagementDefaultLimit() and nvmlDeviceSetPowerManagementLimit().

• Added nvmlDeviceGetInforomImageVersion().

• Expanded nvmlDeviceGetUUID() to support all CUDA capable GPUs.

• Deprecated nvmlDeviceGetDetailedEccErrors() in favor of nvmlDeviceGetMemoryErrorCounter().

• Added NVML_MEMORY_LOCATION_TEXTURE_MEMORY to support reporting of texture memory error counters.

• Added nvmlDeviceGetCurrentClocksThrottleReasons() and nvmlDeviceGetSupportedClocksThrottleReasons().

• NVML_CLOCK_SM is now also reported on supported Kepler devices.

• Dropped support for GT200 based Tesla brand GPUs: C1060, M1060, S1070.

Changes between NVML v2.285 and v3.295

The following new functionality is exposed on NVIDIA display drivers version 295 or later.

• Deprecated nvmlDeviceGetHandleBySerial() in favor of newly added nvmlDeviceGetHandleByUUID().

• Marked the input parameters of nvmlDeviceGetHandleBySerial(), nvmlDeviceGetHandleByUUID() and nvmlDeviceGetHandleByPciBusId as const.

• Added nvmlDeviceOnSameBoard().

• Added nvmlConstants defines.

• Added nvmlDeviceGetMaxPcieLinkGeneration(), nvmlDeviceGetMaxPcieLinkWidth(), nvmlDeviceGetCurrPcieLinkGeneration(),:cpp:func:nvmlDeviceGetCurrPcieLinkWidth.

• Format change of nvmlDeviceGetUUID() output to match the UUID standard. This function will return a different value.

• nvmlDeviceGetDetailedEccErrors() will report zero for unsupported ECC error counters when a subset of ECC error counters are supported.

Changes between NVML v1.0 and v2.285

The following new functionality is exposed on NVIDIA display drivers version 285 or later.

• Added possibility to query separately current and pending driver model with nvmlDeviceGetDriverModel.

  • Fixed error message when querying pending driver model with nvmlDeviceGetDriverModel on non-Admin user account.

  • Fixed problem when nvmlDeviceGetDisplayMode() wouldn’t update.

• Added API nvmlDeviceGetVbiosVersion() function to report VBIOS version.

• Added pciSubSystemId to nvmlPciInfo_t struct.

• Added API nvmlErrorString() function to convert error code to string.

• Updated docs to indicate we support M2075 and C2075.

• Added API nvmlSystemGetHicVersion() function to report HIC firmware version.

• Added NVML versioning support

  • Functions that changed API and/or size of structs have appended versioning suffix (e.g., nvmlDeviceGetPciInfo_v2). Appropriate C defines have been added that map old function names to the newer version of the function.

• Added support for concurrent library usage by multiple libraries.

• Added API nvmlDeviceGetMaxClockInfo() function for reporting device’s clock limits.

• Added new error code NVML_ERROR_DRIVER_NOT_LOADED used by nvmlInit.

• Extended nvmlPciInfo_t struct with new field: sub system id.

• Added NVML support on Windows guest account.

• Changed format of pciBusId string (to XXXX:XX:XX.X) of nvmlPciInfo_t.

• Parsing of busId in nvmlDeviceGetHandleByPciBusId is less restrictive. You can pass 0:2:0.0 or 0000:02:00 and other variations.

• Added API for events waiting for GPU events (Linux only) see docs of nvmlEvents.

• Added API nvmlDeviceGetComputeRunningProcesses_v2 and nvmlSystemGetProcessName() functions for looking up currently running compute applications.

• Deprecated nvmlDeviceGetPowerState() in favor of nvmlDeviceGetPerformanceState().