CUDAErrorString retrieves an enum value from the enum constants string name.

CUDAErrorValues returns all values of the enum.

CUDAMemcpyKindString retrieves an enum value from the enum constants string name.

CUDAMemcpyKindValues returns all values of the enum.

CUdevice_attributeString retrieves an enum value from the enum constants string name.

CUdevice_attributeValues returns all values of the enum.

CUpti_ActivityAttributeString retrieves an enum value from the enum constants string name.

CUpti_ActivityAttributeValues returns all values of the enum.

CUpti_ActivityComputeApiKindString retrieves an enum value from the enum constants string name.

CUpti_ActivityComputeApiKindValues returns all values of the enum.

CUpti_ActivityEnvironmentKindString retrieves an enum value from the enum constants string name.

CUpti_ActivityEnvironmentKindValues returns all values of the enum.

CUpti_ActivityFlagString retrieves an enum value from the enum constants string name.

CUpti_ActivityFlagValues returns all values of the enum.

CUpti_ActivityInstructionClassString retrieves an enum value from the enum constants string name.

CUpti_ActivityInstructionClassValues returns all values of the enum.

CUpti_ActivityKindString retrieves an enum value from the enum constants string name.

CUpti_ActivityKindValues returns all values of the enum.

CUpti_ActivityMemcpyKindString retrieves an enum value from the enum constants string name.

CUpti_ActivityMemcpyKindValues returns all values of the enum.

CUpti_ActivityMemoryKindString retrieves an enum value from the enum constants string name.

CUpti_ActivityMemoryKindValues returns all values of the enum.

CUpti_ActivityObjectKindString retrieves an enum value from the enum constants string name.

CUpti_ActivityObjectKindValues returns all values of the enum.

CUpti_ActivityOverheadKindString retrieves an enum value from the enum constants string name.

CUpti_ActivityOverheadKindValues returns all values of the enum.

CUpti_ActivityPartitionedGlobalCacheConfigString retrieves an enum value from the enum constants string name.

CUpti_ActivityPartitionedGlobalCacheConfigValues returns all values of the enum.

CUpti_ActivityPCSamplingPeriodString retrieves an enum value from the enum constants string name.

CUpti_ActivityPCSamplingPeriodValues returns all values of the enum.

CUpti_ActivityPCSamplingStallReasonString retrieves an enum value from the enum constants string name.

CUpti_ActivityPCSamplingStallReasonValues returns all values of the enum.

CUpti_ActivityPreemptionKindString retrieves an enum value from the enum constants string name.

CUpti_ActivityPreemptionKindValues returns all values of the enum.

CUpti_ActivityStreamFlagString retrieves an enum value from the enum constants string name.

CUpti_ActivityStreamFlagValues returns all values of the enum.

CUpti_ActivitySynchronizationTypeString retrieves an enum value from the enum constants string name.

CUpti_ActivitySynchronizationTypeValues returns all values of the enum.

CUpti_ActivityThreadIdTypeString retrieves an enum value from the enum constants string name.

CUpti_ActivityThreadIdTypeValues returns all values of the enum.

CUpti_ActivityUnifiedMemoryAccessTypeString retrieves an enum value from the enum constants string name.

CUpti_ActivityUnifiedMemoryAccessTypeValues returns all values of the enum.

CUpti_ActivityUnifiedMemoryCounterKindString retrieves an enum value from the enum constants string name.

CUpti_ActivityUnifiedMemoryCounterKindValues returns all values of the enum.

CUpti_ActivityUnifiedMemoryCounterScopeString retrieves an enum value from the enum constants string name.

CUpti_ActivityUnifiedMemoryCounterScopeValues returns all values of the enum.

CUpti_ActivityUnifiedMemoryMigrationCauseString retrieves an enum value from the enum constants string name.

CUpti_ActivityUnifiedMemoryMigrationCauseValues returns all values of the enum.

CUpti_ActivityUnifiedMemoryRemoteMapCauseString retrieves an enum value from the enum constants string name.

CUpti_ActivityUnifiedMemoryRemoteMapCauseValues returns all values of the enum.

CUpti_ApiCallbackSiteString retrieves an enum value from the enum constants string name.

CUpti_ApiCallbackSiteValues returns all values of the enum.

CUpti_CallbackDomainString retrieves an enum value from the enum constants string name.

CUpti_CallbackDomainValues returns all values of the enum.

CUpti_CallbackIdResourceString retrieves an enum value from the enum constants string name.

CUpti_CallbackIdResourceValues returns all values of the enum.

CUpti_CallbackIdSyncString retrieves an enum value from the enum constants string name.

CUpti_CallbackIdSyncValues returns all values of the enum.

CUpti_DeviceAttributeDeviceClassString retrieves an enum value from the enum constants string name.

CUpti_DeviceAttributeDeviceClassValues returns all values of the enum.

CUpti_DeviceAttributeString retrieves an enum value from the enum constants string name.

CUpti_DeviceAttributeValues returns all values of the enum.

CUpti_DevTypeString retrieves an enum value from the enum constants string name.

CUpti_DevTypeValues returns all values of the enum.

CUpti_driver_api_trace_cbidString retrieves an enum value from the enum constants string name.

CUpti_driver_api_trace_cbidValues returns all values of the enum.

CUpti_EnvironmentClocksThrottleReasonString retrieves an enum value from the enum constants string name.

CUpti_EnvironmentClocksThrottleReasonValues returns all values of the enum.

CUpti_EventAttributeString retrieves an enum value from the enum constants string name.

CUpti_EventAttributeValues returns all values of the enum.

CUpti_EventCategoryString retrieves an enum value from the enum constants string name.

CUpti_EventCategoryValues returns all values of the enum.

CUpti_EventCollectionMethodString retrieves an enum value from the enum constants string name.

CUpti_EventCollectionMethodValues returns all values of the enum.

CUpti_EventCollectionModeString retrieves an enum value from the enum constants string name.

CUpti_EventCollectionModeValues returns all values of the enum.

CUpti_EventDomainAttributeString retrieves an enum value from the enum constants string name.

CUpti_EventDomainAttributeValues returns all values of the enum.

CUpti_EventGroupAttributeString retrieves an enum value from the enum constants string name.

CUpti_EventGroupAttributeValues returns all values of the enum.

CUpti_LinkFlagString retrieves an enum value from the enum constants string name.

CUpti_LinkFlagValues returns all values of the enum.

CUpti_MetricAttributeString retrieves an enum value from the enum constants string name.

CUpti_MetricAttributeValues returns all values of the enum.

CUpti_MetricCategoryString retrieves an enum value from the enum constants string name.

CUpti_MetricCategoryValues returns all values of the enum.

CUpti_MetricEvaluationModeString retrieves an enum value from the enum constants string name.

CUpti_MetricEvaluationModeValues returns all values of the enum.

CUpti_MetricPropertyDeviceClassString retrieves an enum value from the enum constants string name.

CUpti_MetricPropertyDeviceClassValues returns all values of the enum.

CUpti_MetricPropertyIDString retrieves an enum value from the enum constants string name.

CUpti_MetricPropertyIDValues returns all values of the enum.

CUpti_MetricValueKindString retrieves an enum value from the enum constants string name.

CUpti_MetricValueKindValues returns all values of the enum.

CUpti_MetricValueUtilizationLevelString retrieves an enum value from the enum constants string name.

CUpti_MetricValueUtilizationLevelValues returns all values of the enum.

CUpti_nvtx_api_trace_cbidString retrieves an enum value from the enum constants string name.

CUpti_nvtx_api_trace_cbidValues returns all values of the enum.

CUpti_ReadEventFlagsString retrieves an enum value from the enum constants string name.

CUpti_ReadEventFlagsValues returns all values of the enum.

CUPTI_RUNTIME_TRACE_CBIDString retrieves an enum value from the enum constants string name.

CUPTI_RUNTIME_TRACE_CBIDValues returns all values of the enum.

CUptiResultString retrieves an enum value from the enum constants string name.

CUptiResultValues returns all values of the enum.

CUresultString retrieves an enum value from the enum constants string name.

CUresultValues returns all values of the enum.

*< Number of asynchronous engines */.

*< Device can map host memory into CUDA address space */.

*< Deprecated do not use.

*< 64-bit operations are supported in ::cuStreamBatchMemOp and related APIs.

*< Device can access host registered memory at the same virtual address as the CPU */.

*< ::cuStreamBatchMemOp and related APIs are supported.

*< ::CU_STREAM_WAIT_VALUE_NOR is supported.

*< Typical clock frequency in kilohertz */.

*< Major compute capability version number */.

*< Minor compute capability version number */.

*< Compute mode (See ::CUcomputemode for details) */.

*< Device supports compute preemption.

*< Device can possibly execute multiple kernels concurrently */.

*< Device can coherently access managed memory concurrently with the CPU */.

*< Device supports launching cooperative kernels via ::cuLaunchCooperativeKernel */.

*< Device can participate in cooperative kernels launched via ::cuLaunchCooperativeKernelMultiDevice */.

*< Device has ECC support enabled */.

*< Device supports caching globals in L1 */.

*< Global memory bus width in bits */.

*< Device can possibly copy memory and execute a kernel concurrently.

*< Link between the device and the host supports native atomic operations (this is a placeholder attribute and is not supported on any current hardware)*/.

*< Device is integrated with host memory */.

*< Specifies whether there is a run time limit on kernels */.

*< Size of L2 cache in bytes */.

*< Device supports caching locals in L1 */.

*< Device can allocate managed memory on this system */.

*< Maximum block dimension X */.

*< Maximum block dimension Y */.

*< Maximum block dimension Z */.

*< Maximum grid dimension X */.

*< Maximum grid dimension Y */.

*< Maximum grid dimension Z */.

*< Maximum pitch in bytes allowed by memory copies */.

*< Maximum number of 32-bit registers available per block */.

*< Maximum number of 32-bit registers available per multiprocessor */.

*< Maximum shared memory available per block in bytes */.

*< Maximum optin shared memory per block */.

*< Maximum shared memory available per multiprocessor in bytes */.

*< Maximum number of threads per block */.

*< Maximum resident threads per multiprocessor */.

*< Maximum layers in a 1D layered surface */.

*< Maximum 1D layered surface width */.

*< Maximum 1D surface width */.

*< Maximum 2D surface height */.

*< Maximum 2D layered surface height */.

*< Maximum layers in a 2D layered surface */.

*< Maximum 2D layered surface width */.

*< Maximum 2D surface width */.

*< Maximum 3D surface depth */.

*< Maximum 3D surface height */.

*< Maximum 3D surface width */.

*< Maximum layers in a cubemap layered surface */.

*< Maximum cubemap layered surface width */.

*< Maximum cubemap surface width */.

*< Maximum layers in a 1D layered texture */.

*< Maximum 1D layered texture width */.

*< Maximum 1D linear texture width */.

*< Maximum mipmapped 1D texture width */.

*< Maximum 1D texture width */.

*< Deprecated use CU_DEVICE_ATTRIBUTE_MAXIMUM_TEXTURE2D_LAYERED_HEIGHT */.

*< Deprecated use CU_DEVICE_ATTRIBUTE_MAXIMUM_TEXTURE2D_LAYERED_LAYERS */.

*< Deprecated use CU_DEVICE_ATTRIBUTE_MAXIMUM_TEXTURE2D_LAYERED_WIDTH */.

*< Maximum 2D texture height if CUDA_ARRAY3D_TEXTURE_GATHER is set */.

*< Maximum 2D texture width if CUDA_ARRAY3D_TEXTURE_GATHER is set */.

*< Maximum 2D texture height */.

*< Maximum 2D layered texture height */.

*< Maximum layers in a 2D layered texture */.

*< Maximum 2D layered texture width */.

*< Maximum 2D linear texture height */.

*< Maximum 2D linear texture pitch in bytes */.

*< Maximum 2D linear texture width */.

*< Maximum mipmapped 2D texture height */.

*< Maximum mipmapped 2D texture width */.

*< Maximum 2D texture width */.

*< Maximum 3D texture depth */.

*< Alternate maximum 3D texture depth */.

*< Maximum 3D texture height */.

*< Alternate maximum 3D texture height */.

*< Maximum 3D texture width */.

*< Alternate maximum 3D texture width */.

*< Maximum layers in a cubemap layered texture */.

*< Maximum cubemap layered texture width/height */.

*< Maximum cubemap texture width/height */.

*< Peak memory clock frequency in kilohertz */.

*< Device is on a multi-GPU board */.

*< Unique id for a group of devices on the same multi-GPU board */.

*< Number of multiprocessors on device */.

*< Device supports coherently accessing pageable memory without calling cudaHostRegister on it */.

*< PCI bus ID of the device */.

*< PCI device ID of the device */.

*< PCI domain ID of the device */.

*< Deprecated use CU_DEVICE_ATTRIBUTE_MAX_REGISTERS_PER_BLOCK */.

*< Deprecated use CU_DEVICE_ATTRIBUTE_MAX_SHARED_MEMORY_PER_BLOCK */.

*< Ratio of single precision performance (in floating-point operations per second) to double precision performance */.

*< Device supports stream priorities */.

*< Alignment requirement for surfaces */.

*< Device is using TCC driver model */.

*< Alignment requirement for textures */.

*< Pitch alignment requirement for textures */.

*< Memory available on device for __constant__ variables in a CUDA C kernel in bytes */.

*< Device shares a unified address space with the host */.

*< Warp size in threads */.

* * This indicates that a resource has already been acquired.

* * This indicates that the resource is already mapped.

* * This indicates that the specified array is currently mapped and thus * cannot be destroyed.

* * A device-side assert triggered during kernel execution.

* * This indicated that the context being supplied as a parameter to the * API call was already the active context.

* * This indicates that the ::CUcontext passed to the API call can * only be bound to a single CPU thread at a time but is already * bound to a CPU thread.

* * This error indicates that the context current to the calling thread * has been destroyed using ::cuCtxDestroy or is a primary context which * has not yet been initialized.

* * This error indicates that the number of blocks launched per grid for a kernel that was * launched via either ::cuLaunchCooperativeKernel or ::cuLaunchCooperativeKernelMultiDevice * exceeds the maximum number of blocks as allowed by ::cuOccupancyMaxActiveBlocksPerMultiprocessor * or ::cuOccupancyMaxActiveBlocksPerMultiprocessorWithFlags times the number of multiprocessors * as specified by the device attribute ::CU_DEVICE_ATTRIBUTE_MULTIPROCESSOR_COUNT.

* * This indicates that the CUDA driver is in the process of shutting down.

* * This indicates that an uncorrectable ECC error was detected during * execution.

* * This indicates that the file specified was not found.

* * While executing a kernel the device encountered a stack error.

* * This error indicates that the memory range passed to ::cuMemHostRegister() * has already been registered.

* * This error indicates that the pointer passed to ::cuMemHostUnregister() * does not correspond to any currently registered memory region.

* * While executing a kernel the device encountered a * load or store instruction on an invalid memory address.

* * While executing a kernel the device encountered an illegal instruction.

* * While executing a kernel the device encountered an instruction * which can only operate on memory locations in certain address spaces * (global shared or local) but was supplied a memory address not * belonging to an allowed address space.

* * This most frequently indicates that there is no context bound to the * current thread.

* * This indicates that the device ordinal supplied by the user does not * correspond to a valid CUDA device.

* * This indicates an error with OpenGL or DirectX context.

* * This indicates that a resource handle passed to the API call was not * valid.

* * This indicates that the device kernel image is invalid.

* * While executing a kernel the device program counter wrapped its address space.

* * This indicates that a PTX JIT compilation failed.

* * This indicates that the device kernel source is invalid.

* * This indicates that one or more of the parameters passed to the API call * is not within an acceptable range of values.

* * This indicates that the PTX JIT compiler library was not found.

* * An exception occurred on the device while executing a kernel.

* * This error indicates a kernel launch that uses an incompatible texturing * mode.

* * This indicates that a launch did not occur because it did not have * appropriate resources.

* * This indicates that the device kernel took too long to execute.

* * This indicates that a map or register operation has failed.

* * While executing a kernel the device encountered a load or store instruction * on a memory address which is not aligned.

* * This indicates that there is no kernel image available that is suitable * for the device.

* * This indicates that no CUDA-capable devices were detected by the installed * CUDA driver.

* * This indicates that a named symbol was not found.

* * This indicates that the CUDA driver has not been initialized with * ::cuInit() or that initialization has failed.

* * This indicates that a resource is not mapped.

* * This indicates that a mapped resource is not available for access as an * array.

* * This indicates that a mapped resource is not available for access as a * pointer.

* * This error indicates that the attempted operation is not permitted.

* * This indicates that asynchronous operations issued previously have not * completed yet.

* * This error indicates that the attempted operation is not supported * on the current system or device.

* * This indicates that an uncorrectable NVLink error was detected during the * execution.

* * This indicates that an OS call failed.

* * The API call failed because it was unable to allocate enough memory to * perform the requested operation.

* * This error indicates that a call to ::cuCtxEnablePeerAccess() is * trying to re-enable peer access to a context which has already * had peer access to it enabled.

* * This error indicates that ::cuCtxDisablePeerAccess() is * trying to disable peer access which has not been enabled yet * via ::cuCtxEnablePeerAccess().

* * This indicates that peer access is not supported across the given * devices.

* * This error indicates that the primary context for the specified device * has already been initialized.

* * \deprecated * This error return is deprecated as of CUDA 5.0.

* * \deprecated * This error return is deprecated as of CUDA 5.0.

* * This indicates profiler is not initialized for this run.

* * \deprecated * This error return is deprecated as of CUDA 5.0.

* * This indicates that initialization of a shared object failed.

* * This indicates that a link to a shared object failed to resolve.

* * This error indicates that the hardware resources required to enable * peer access have been exhausted for one or more of the devices * passed to ::cuCtxEnablePeerAccess().

* * This indicates that an unknown internal error has occurred.

* * This indicates that an unmap or unregister operation has failed.

* * This indicates that the ::CUlimit passed to the API call is not * supported by the active device.

* * The API call returned with no errors.

* * This indicated that the user has taken the address of a constant variable * which was forbidden up until the CUDA 3.1 release.

* * Any unhandled CUDA driver error is added to this value and returned via * the runtime.

* * An assert triggered in device code during kernel execution.

* * This error indicates that the number of blocks launched per grid for a kernel that was * launched via either ::CUDALaunchCooperativeKernel or ::CUDALaunchCooperativeKernelMultiDevice * exceeds the maximum number of blocks as allowed by ::CUDAOccupancyMaxActiveBlocksPerMultiprocessor * or ::CUDAOccupancyMaxActiveBlocksPerMultiprocessorWithFlags times the number of multiprocessors * as specified by the device attribute ::CUDADevAttrMultiProcessorCount.

* * This indicates that a CUDA Runtime API call cannot be executed because * it is being called during process shut down at a point in time after * CUDA driver has been unloaded.

* * This indicates that a call tried to access an exclusive-thread device that * is already in use by a different thread.

* * This indicates that all CUDA devices are busy or unavailable at the current * time.

* * This indicates that multiple surfaces (across separate CUDA source * files in the application) share the same string name.

* * This indicates that multiple textures (across separate CUDA source * files in the application) share the same string name.

* * This indicates that multiple global or constant variables (across separate * CUDA source files in the application) share the same string name.

* * This indicates that an uncorrectable ECC error was detected during * execution.

* * Device encountered an error in the call stack during kernel execution * possibly due to stack corruption or exceeding the stack size limit.

* * This error indicates that the memory range passed to ::CUDAHostRegister() * has already been registered.

* * This error indicates that the pointer passed to ::CUDAHostUnregister() * does not correspond to any currently registered memory region.

* * The device encountered a load or store instruction on an invalid memory address.

* * The device encountered an illegal instruction during kernel execution * This leaves the process in an inconsistent state and any further CUDA work * will return the same error.

* * This indicates that the current context is not compatible with this * the CUDA Runtime.

* * The API call failed because the CUDA driver and runtime could not be * initialized.

* * This indicates that the installed NVIDIA CUDA driver is older than the * CUDA runtime library.

* * While executing a kernel the device encountered an instruction * which can only operate on memory locations in certain address spaces * (global shared or local) but was supplied a memory address not * belonging to an allowed address space.

* * This indicates that the channel descriptor passed to the API call is not * valid.

* * This indicates that a kernel launch is requesting resources that can * never be satisfied by the current device.

* * This indicates that the device ordinal supplied by the user does not * correspond to a valid CUDA device.

* * The requested device function does not exist or is not compiled for the * proper device architecture.

* * This indicates that at least one device pointer passed to the API call is * not a valid device pointer.

* * This indicates that a non-float texture was being accessed with linear * filtering.

* * This indicates an error with the OpenGL or DirectX context.

* * This indicates that at least one host pointer passed to the API call is * not a valid host pointer.

* * This indicates that the device kernel image is invalid.

* * This indicates that the direction of the memcpy passed to the API call is * not one of the types specified by ::CUDAMemcpyKind.

* * This indicates that an attempt was made to read a non-float texture as a * normalized float.

* * The device encountered an invalid program counter.

* * This indicates that one or more of the pitch-related parameters passed * to the API call is not within the acceptable range for pitch.

* * A PTX compilation failed.

* * This indicates that a resource handle passed to the API call was not * valid.

* * This indicates that the surface passed to the API call is not a valid * surface.

* * This indicates that the symbol name/identifier passed to the API call * is not a valid name or identifier.

* * This indicates that the texture passed to the API call is not a valid * texture.

* * This indicates that the texture binding is not valid.

* * This indicates that one or more of the parameters passed to the API call * is not within an acceptable range of values.

* * This indicates that the PTX JIT compiler library was not found.

* * An exception occurred on the device while executing a kernel.

* * This error indicates that a grid launch did not occur because the kernel * uses file-scoped surfaces which are unsupported by the device runtime.

* * This error indicates that a grid launch did not occur because the kernel * uses file-scoped textures which are unsupported by the device runtime.

* * This error indicates that a device runtime grid launch did not occur * because the depth of the child grid would exceed the maximum supported * number of nested grid launches.

* * This indicates that a launch did not occur because it did not have * appropriate resources.

* * This error indicates that a device runtime grid launch failed because * the launch would exceed the limit ::CUDALimitDevRuntimePendingLaunchCount.

* * This indicates that the device kernel took too long to execute.

* * This indicates that the buffer object could not be mapped.

* * The API call failed because it was unable to allocate enough memory to * perform the requested operation.

* * This indicated that an emulated device pointer exceeded the 32-bit address * range.

* * The device encountered a load or store instruction * on a memory address which is not aligned.

* * The device function being invoked (usually via ::CUDALaunchKernel()) was not * previously configured via the ::CUDAConfigureCall() function.

* * Mixing of device and device emulation code was not allowed.

* * This indicates that no CUDA-capable devices were detected by the installed * CUDA driver.

* * This indicates that there is no kernel image available that is suitable * for the device.

* * This error indicates the attempted operation is not permitted.

* * This indicates that asynchronous operations issued previously have not * completed yet.

* * This error indicates the attempted operation is not supported * on the current system or device.

* * This indicates that the API call is not yet implemented.

* * This indicates that an uncorrectable NVLink error was detected during the * execution.

* * This error indicates that an OS call failed.

* * This error indicates that a call to ::CUDADeviceEnablePeerAccess() is * trying to re-enable peer addressing on from a context which has already * had peer addressing enabled.

* * This error indicates that ::CUDADeviceDisablePeerAccess() is trying to * disable peer addressing which has not been enabled yet via * ::CUDADeviceEnablePeerAccess().

* * This error indicates that P2P access is not supported across the given * devices.

* * This indicated that a previous kernel launch failed.

* * \deprecated * This error return is deprecated as of CUDA 5.0.

* * \deprecated * This error return is deprecated as of CUDA 5.0.

* * This indicates profiler is not initialized for this run.

* * \deprecated * This error return is deprecated as of CUDA 5.0.

* * This indicates that the user has called ::CUDASetValidDevices() * ::CUDASetDeviceFlags() ::CUDAD3D9SetDirect3DDevice() * ::CUDAD3D10SetDirect3DDevice ::CUDAD3D11SetDirect3DDevice() or * ::CUDAVDPAUSetVDPAUDevice() after initializing the CUDA runtime by * calling non-device management operations (allocating memory and * launching kernels are examples of non-device management operations).

* * This indicates that initialization of a shared object failed.

* * This indicates that a link to a shared object failed to resolve.

* * This indicates an internal startup failure in the CUDA runtime.

* * This error indicates that a call to ::CUDADeviceSynchronize made from * the device runtime failed because the call was made at grid depth greater * than than either the default (2 levels of grids) or user specified device * limit ::CUDALimitDevRuntimeSyncDepth.

* * This indicated that a synchronization operation had failed.

* * This indicated that a texture fetch was not able to be performed.

* * This indicated that a texture was not bound for access.

* * This error indicates that the hardware resources required to enable * peer access have been exhausted for one or more of the devices * passed to ::CUDAEnablePeerAccess().

* * This indicates that an unknown internal error has occurred.

* * This indicates that the buffer object could not be unmapped.

* * This indicates that the ::CUDALimit passed to the API call is not * supported by the active device.

* * The API call returned with no errors.

* * The maximum number of memory buffers per context.

* * The device memory size (in bytes) reserved for storing profiling * data for non-CDP operations especially for concurrent kernel * tracing for each buffer on a context.

* * The device memory size (in bytes) reserved for storing profiling * data for CDP operations for each buffer on a context.

* * The maximum number of profiling semaphore pools per context.

* * The profiling semaphore pool size reserved for storing profiling * data for serialized kernels and memory operations for each context.

* * The compute APIs are for CUDA.

* * The compute APIs are for CUDA running * in MPS (Multi-Process Service) environment.

* * The compute API is not known.

* * The environment data is related to cooling.

* * The environment data is related to power.

* * The environment data is related to speed.

* * The environment data is related to temperature.

* * Unknown data.

* * Indicates if the activity represents a CUdevice_attribute value * or a CUpti_DeviceAttribute value.

* * Indicates the activity represents a device that supports * concurrent kernel execution.

* * When calling cuptiActivityFlushAll this flag * can be set to force CUPTI to flush all records in the buffer whether * finished or not */.

* * If this bit in flag is set the load access was cached else it is * uncached.

* * If bit in this flag is set the access was load else it is a * store access.

* * The number of bytes requested by each thread * Valid for CUpti_ActivityGlobalAccess3.

* * The mask for the instruction class \ref CUpti_ActivityInstructionClass * Valid for CUpti_ActivityInstructionExecution and * CUpti_ActivityInstructionCorrelation */.

* * If this bit in flag is set the source level metric value couldn't be * calculated.

* * Indicates the activity represents a marker that specifies a color * in alpha-red-green-blue format.

* * Indicates the activity represents a marker that does not specify * a color.

* * Indicates the activity represents a region end marker.

* * Indicates the activity represents an instantaneous marker.

* * Indicates the activity represents a region start marker.

* * Indicates the activity represents an attempt to acquire a user * defined synchronization object.

* * Indicates the activity represents failure in acquiring the * user defined synchronization object.

* * Indicates the activity represents success in acquiring the * user defined synchronization object.

* * Indicates the activity represents releasing a reservation on * user defined synchronization object.

* * Indicates the activity represents an asynchronous memcpy * operation.

* * Indicates the activity represents an asynchronous memset * operation.

* * If this bit in flag is set the metric value overflowed.

* * If this bit in flag is set the metric value couldn't be * calculated.

* * Indicates the activity record has no flags.

* * If bit in this flag is set the access was load else it is a * store access.

* * The number of bytes requested by each thread * Valid for CUpti_ActivitySharedAccess.

* * Indicates the activity represents thrashing in CPU.

* * Indicates the activity represents page throttling in CPU.

* * Represents a barrier instruction.

* * Represents a bit conversion operation.

* * Represents a constant load instruction.

* * Represents a control flow instruction.

* * Represents a 16 bit floating point operation.

* * Represents a 32 bit floating point operation.

* * Represents a 64 bit floating point operation.

* * Represents a generic load-store instruction.

* * Represents a global load-store instruction.

* * Represents a global atomic instruction.

* * Represents an integer operation.

* * Represents a inter-thread communication instruction.

* * Represents a local load-store instruction.

* * Represents some miscellaneous instructions which do not fit in the above classification.

* * Represents a shared load-store instruction.

* * Represents a shared atomic instruction.

* * Represents a surface load-store instruction.

* * Represents a surface atomic instruction.

* * Represents a texture load-store instruction.

* * The instruction class is not known.

* * Results for source-level branch.

* * A CDP (CUDA Dynamic Parallel) kernel executing on the GPU.

* * A (potentially concurrent) kernel executing on the GPU.

* * Information about a context.

* * Information about a CUDA event.

* * Information about a device.

* * A device attribute value.

* * A CUDA driver API function execution.

* * Environment activity records indicating power clock thermal * etc.

* * An event value.

* * An event value associated with a specific event domain * instance.

* * Records for correlation of different programming APIs.

* * Device global/function record.

* * Results for source-level global acccess.

* * Instantaneous Event information.

* * Instantaneous Event information for a specific event * domain instance.

* * Instantaneous Metric information * The corresponding activity record structure is \ref * CUpti_ActivityInstantaneousMetric.

* * Instantaneous Metric information for a specific metric * domain instance.

* * SASS/Source line-by-line correlation record.

* * Results for source-level instruction execution.

* * The activity record is invalid.

* * A kernel executing on the GPU.

* * Instantaneous start or end marker.

* * Extended optional data about a marker.

* * A host<->host host<->device or device<->device memory copy.

* * A peer to peer memory copy.

* Memory activity tracking allocation and freeing of the memory * The corresponding activity record structure is \ref * CUpti_ActivityMemory.

* * A memory set executing on the GPU.

* * A metric value.

* * A metric value associated with a specific metric domain * instance.

* * CUDA Module record.

* * Thread device context etc.

* * NVLink information.

* * OpenACC data events.

* * OpenACC launch events.

* * OpenACC other events.

* * Overhead activity records.

* * Enable PC sampling for kernels.

* * Summary information about PC sampling records.

* * Preemption activity record indicating a preemption of a CDP (CUDA * Dynamic Parallel) kernel executing on the GPU.

* * A CUDA runtime API function execution.

* * Results for source-level shared acccess.

* * Source information about source level result.

* * Information about a CUDA stream.

* * Records for synchronization management.

* * Unified Memory counter record.

* * A device array to device array memory copy.

* * A device array to device memory copy.

* * A device array to host memory copy.

* * A device to device array memory copy.

* * A device to device memory copy on the same device.

* * A device to host memory copy.

* * A host to device array memory copy.

* * A host to device memory copy.

* * A host to host memory copy.

* * A peer to peer memory copy across different devices.

* * The memory copy kind is not known.

* * The memory is an array.

* * The memory is on the device.

* * The memory is device static */.

* * The memory is managed */.

* * The memory is managed static */.

* * The memory is pageable.

* * The memory is pinned.

* * The memory kind is unknown.

* * A context.

* * A device.

* * A process.

* * A stream.

* * A thread.

* * The object kind is not known.

* * Activity buffer flush overhead.

* * CUPTI instrumentation overhead.

* * CUPTI resource creation and destruction overhead.

* * Compiler(JIT) overhead.

* * The overhead kind is not known.

* * Partitioned global cache not supported.

* * Partitioned global cache config off.

* * Partitioned global cache config on.

* * Partitioned global cache config unknown.

* * Sampling period in higher range.

* * The PC sampling period is not set.

* * Sampling period in lower range.

* * Maximum sampling period available on the device.

* * Medium sampling period.

* * Minimum sampling period available on the device.

* * Warp is blocked waiting for __constant__ memory and immediate memory access to complete.

* * Instruction is waiting on an arithmatic dependency */.

* * Warp is blocked because next instruction is not yet available * because of instruction cache miss or because of branching effects */.

* * Invalid reason */.

* * Warp is blocked because it is waiting for a memory access to complete.

* * Warp is blocked because there are too many pending memory operations.

* * No stall instruction is selected for issue */.

* * Warp was ready to issue but some other warp issued instead.

* * Miscellaneous reasons */.

* * Compute operation cannot be performed due to the required resources not * being available.

* * Sleeping.

* * Warp is blocked as it is waiting at __syncthreads() or at memory barrier.

* * Texture sub-system is fully utilized or has too many outstanding requests.

* * Preemption to restore CDP block.

* * Preemption to save CDP block.

* * The preemption kind is not known.

* * Default stream.

* * Non-blocking stream.

* * Null stream.

* * Unknown data.

* * Stream create Mask */.

* * Context synchronize API.

* * Event synchronize API.

* * Stream synchronize API.

* * Stream wait event API.

* * Unknown data.

* * Default type * Windows uses API GetCurrentThreadId() * Linux/Mac/Android/QNX use POSIX pthread API pthread_self() */.

* * This type is based on the system API available on the underlying platform * and thread-id obtained is supposed to be unique for the process lifetime.

* * The page fault was triggered by atomic memory instruction */.

* * The page fault was triggered by memory prefetch operation */.

* * The page fault was triggered by read memory instruction */.

* * The unified memory access type is not known */.

* * The page fault was triggered by write memory instruction */.

* * Number of bytes transferred from one device to another device.

* * Number of bytes transfered from device to host */.

* * Number of bytes transfered from host to device */.

* * Number of CPU page faults this is only supported on 64 bit * Linux and Mac platforms */.

* * Number of GPU page faults this is only supported on devices with * compute capability 6.0 and higher and 64 bit Linux platforms */.

* * In case throttling does not help the driver tries to pin the memory * to a processor for a specific period of time.

* * Thrashing occurs when data is frequently accessed by * multiple processors and has to be constantly migrated around * to achieve data locality.

* * Throttling is a prevention technique used by the driver to avoid * further thrashing.

* * The unified memory counter kind is not known.

* * Collect unified memory counter for single process across all devices */.

* * Collect unified memory counter for single process on one device */.

* * The unified memory counter scope is not known.

* * The unified memory migrated to guarantee data coherence * e.g.

* * The unified memory migrated to the CPU because it was evicted to make * room for another block of memory on the GPU */.

* * The unified memory was speculatively migrated by the UVM driver * before being accessed by the destination processor to improve * performance */.

* * The unified memory migration cause is not known */.

* * The unified memory migrated due to an explicit call from * the user e.g.

* * Mapping to remote memory was added to maintain data coherence.

* * Mapping to remote memory was added because there is no more * memory available on the processor and eviction was not * possible */.

* * Mapping to remote memory was added to enforce the hints * specified by the programmer or by performance heuristics of the * UVM driver */.

* * Mapping to remote memory was added to prevent further thrashing */.

* * The cause of mapping to remote memory was unknown */.

* * The callback is at the entry of the API call.

* * The callback is at the exit of the API call.

* * Domain containing callback points for all driver API functions.

* * Invalid domain.

* * Domain containing callback points for NVTX API functions.

* * Domain containing callback points for CUDA resource tracking.

* * Domain containing callback points for all runtime API * functions.

* * Domain containing callback points for CUDA synchronization.

* * A new context has been created.

* * A context is about to be destroyed.

* * The driver has finished initializing.

* * Invalid resource callback ID.

* * A module has been loaded.

* * The current module which is being profiled.

* * A module is about to be unloaded.

* * A new stream has been created.

* * A stream is about to be destroyed.

* * Context synchronization has completed for the context.

* * Invalid synchronize callback ID.

* * Stream synchronization has completed for the stream.

* * Nothing is running on the GPU and the clocks are dropping to idle * state.

* * Hardware slowdown to reduce the clock by a factor of two or more * is engaged.

* * No clock throttling.

* * A software power scaling algorithm is reducing the clocks below * requested clocks.

* * Some unspecified factor is reducing the clocks.

* * Throttle reason is not supported for this GPU.

* * The GPU clocks are limited by a user specified limit.

* * The device type is GPU.

* * The device type is NVLink processing unit in CPU.

* * Get the class for the device.

* * Get the peak double precision flop per cycle.

* * Get the peak half precision flop per cycle.

* * Get the peak single precision flop per cycle.

* * Get global memory bandwidth in Kbytes/sec.

* * Check if Nvlink is present between GPU and CPU.

* * Get theoretical maximum number of instructions per cycle.

* * Get theoretical maximum number of single precision instructions * that can be executed per second.

* * Number of event domain IDs for a device.

* * Number of event IDs for a device.

* * Get number of frame buffers for device.

* * Get number of L2 units.

* * Get the maximum shared memory for the CU_FUNC_CACHE_PREFER_EQUAL * preference.

* * Get the maximum shared memory for the CU_FUNC_CACHE_PREFER_L1 * preference.

* * Get the maximum shared memory for the CU_FUNC_CACHE_PREFER_SHARED * preference.

* * Check if Nvlink is connected to device.

* * Get PCIE generation for device.

* * Get PCIE link rate in Mega bits/sec for device.

* * Get PCIE link width for device.

* * API is not implemented.

* * CUPTI is disabled due to conflicts with other enabled profilers */.

* * An error occurred on the performance monitoring hardware.

* * The performance monitoring hardware is in use by other client.

* * The context is NULL or not valid.

* * The device does not correspond to a valid CUDA device.

* * The event domain id is invalid.

* * The event id is invalid.

* * The event name is invalid.

* * Invalid event value.

* * Invalid handle (internal?).

* * Invalid kind.

* * The metric id is invalid.

* * The metric name is invalid.

* * Invalid metric value.

* * Invalid module.

* * The current operation cannot be performed due to dependency on * other factors.

* * One or more of the parameters is invalid.

* * Invalid stream.

* * The maximum limit is reached.

* * The current operation is not compatible with the current state * of the object */.

* * CUPTI is unable to initialize its connection to the CUDA * driver.

* * The object is not yet ready to perform the requested operation.

* * The attempted operation is not supported on the current * system or device.

* * Unable to allocate enough memory to perform the requested * operation.

* * The output buffer size is not sufficient to return all * requested data.

* * The queue is empty.

* * Unified memory profiling is not supported on the system.

* * Unified memory profiling is not supported on the device */.

* * Unified memory profiling is not supported on a multi-GPU * configuration without P2P support between any pair of devices */.

* * Unified memory profiling is not supported under the * Multi-Process Service (MPS) environment.

* * An unknown internal error has occurred.

* * Category of event.

* * Long description of event.

* * Event name.

* * Short description of event.

* * A cache related event.

* * An instruction related event.

* * A memory related event.

* * A profile-trigger event.

* * Event is collected using software instrumentation.

* * Event is collected using NvLink throughput counter method.

* * Event is collected using a hardware global performance monitor.

* * Event is collected using a hardware SM performance monitor.

* * Events are collected for the entire duration between the * cuptiEventGroupEnable and cuptiEventGroupDisable calls.

* * Events are collected only for the durations of kernel executions * that occur between the cuptiEventGroupEnable and * cuptiEventGroupDisable calls.

* * Collection method used for events contained in the event domain.

* * Number of instances of the domain for which event counts will be * collected.

* * Event domain name.

* * Total number of instances of the domain including instances that * cannot be profiled.

* * The domain to which the event group is bound.

* * Enumerates events in the group.

* * Number of instances of the domain bound to this event group that * will be counted.

* * Number of events in the group.

* * [rw] Profile all the instances of the domain for this * eventgroup.

* * [rw] Reserved for user data.

* * No flags.

* * Is peer to peer access supported by this link.

* * Is peer atomic access supported by this link.

* * Is system memory access supported by this link.

* * Is system memory atomic access supported by this link.

* * Category of the metric.

* * Metric evaluation mode.

* * Long description of metric.

* * Metric name.

* * Short description of metric.

* * Value type of the metric.

* * A cache related metric.

* * An instruction related metric.

* * A memory related metric.

* * A multiprocessor related metric.

* *A Nvlink related metric.

* * A texture related metric.

* * If this bit is set the metric can be profiled over all instances.

* * If this bit is set the metric can be profiled for each instance of the * domain.

* Clock rate for device in KHz.

* The device class.

* Whether ECC support is enabled on the device.

* Peak double precision floating point operations that * can be performed in one cycle by the device.

* Peak half precision floating point operations that * can be performed in one cycle by the device.

* Peak single precision floating point operations that * can be performed in one cycle by the device.

* Number of Frame buffer units for device.

* Global memory bandwidth in KBytes/sec.

* NVLINK Bandwitdh for device.

* GPU Time for kernel in ns.

* Number of L2 units on a device.

* Number of multiprocessors on a device.

* PCIE generation for device.

* PCIE link rate in Mega bits/sec.

* PCIE link width for device.

* Maximum number of warps on a multiprocessor.

* * The metric value is a 64-bit double.

* * The metric value is a 64-bit signed integer.

* * The metric value is a percentage represented by a 64-bit * double.

* * The metric value is a throughput represented by a 64-bit * integer.

* * The metric value is a 64-bit unsigned integer.

* * The metric value is a utilization level as represented by * CUpti_MetricValueUtilizationLevel.

* * No error.