dtfft_interface_cuda_runtime Module

CUDA Runtime Interfaces


Uses

  • module~~dtfft_interface_cuda_runtime~~UsesGraph module~dtfft_interface_cuda_runtime dtfft_interface_cuda_runtime iso_c_binding iso_c_binding module~dtfft_interface_cuda_runtime->iso_c_binding module~dtfft_parameters dtfft_parameters module~dtfft_interface_cuda_runtime->module~dtfft_parameters module~dtfft_utils dtfft_utils module~dtfft_interface_cuda_runtime->module~dtfft_utils module~dtfft_parameters->iso_c_binding iso_fortran_env iso_fortran_env module~dtfft_parameters->iso_fortran_env mpi_f08 mpi_f08 module~dtfft_parameters->mpi_f08 module~dtfft_utils->iso_c_binding module~dtfft_utils->module~dtfft_parameters module~dtfft_utils->iso_fortran_env module~dtfft_errors dtfft_errors module~dtfft_utils->module~dtfft_errors module~dtfft_utils->mpi_f08 module~dtfft_errors->iso_fortran_env

Used by

  • module~~dtfft_interface_cuda_runtime~~UsedByGraph module~dtfft_interface_cuda_runtime dtfft_interface_cuda_runtime module~dtfft_abstract_backend dtfft_abstract_backend module~dtfft_abstract_backend->module~dtfft_interface_cuda_runtime module~dtfft_config dtfft_config module~dtfft_abstract_backend->module~dtfft_config module~dtfft_nvrtc_kernel dtfft_nvrtc_kernel module~dtfft_abstract_backend->module~dtfft_nvrtc_kernel module~dtfft_pencil dtfft_pencil module~dtfft_abstract_backend->module~dtfft_pencil module~dtfft_abstract_transpose_plan dtfft_abstract_transpose_plan module~dtfft_abstract_transpose_plan->module~dtfft_interface_cuda_runtime module~dtfft_abstract_transpose_plan->module~dtfft_abstract_backend module~dtfft_abstract_transpose_plan->module~dtfft_config module~dtfft_abstract_transpose_plan->module~dtfft_nvrtc_kernel module~dtfft_abstract_transpose_plan->module~dtfft_pencil module~dtfft_backend_cufftmp_m dtfft_backend_cufftmp_m module~dtfft_backend_cufftmp_m->module~dtfft_interface_cuda_runtime module~dtfft_backend_cufftmp_m->module~dtfft_abstract_backend module~dtfft_backend_cufftmp_m->module~dtfft_pencil module~dtfft_backend_mpi dtfft_backend_mpi module~dtfft_backend_mpi->module~dtfft_interface_cuda_runtime module~dtfft_backend_mpi->module~dtfft_abstract_backend module~dtfft_backend_nccl_m dtfft_backend_nccl_m module~dtfft_backend_nccl_m->module~dtfft_interface_cuda_runtime module~dtfft_backend_nccl_m->module~dtfft_abstract_backend module~dtfft_config->module~dtfft_interface_cuda_runtime module~dtfft_executor_cufft_m dtfft_executor_cufft_m module~dtfft_executor_cufft_m->module~dtfft_interface_cuda_runtime module~dtfft_executor_cufft_m->module~dtfft_config module~dtfft_abstract_executor dtfft_abstract_executor module~dtfft_executor_cufft_m->module~dtfft_abstract_executor module~dtfft_nvrtc_block_optimizer dtfft_nvrtc_block_optimizer module~dtfft_nvrtc_block_optimizer->module~dtfft_interface_cuda_runtime module~dtfft_nvrtc_block_optimizer->module~dtfft_config module~dtfft_nvrtc_kernel->module~dtfft_interface_cuda_runtime module~dtfft_nvrtc_kernel->module~dtfft_config module~dtfft_nvrtc_kernel->module~dtfft_nvrtc_block_optimizer module~dtfft_nvrtc_kernel_cache dtfft_nvrtc_kernel_cache module~dtfft_nvrtc_kernel->module~dtfft_nvrtc_kernel_cache module~dtfft_nvrtc_kernel_generator dtfft_nvrtc_kernel_generator module~dtfft_nvrtc_kernel->module~dtfft_nvrtc_kernel_generator module~dtfft_nvrtc_kernel_cache->module~dtfft_interface_cuda_runtime module~dtfft_nvrtc_kernel_cache->module~dtfft_config module~dtfft_nvrtc_kernel_generator->module~dtfft_interface_cuda_runtime module~dtfft_nvrtc_kernel_generator->module~dtfft_nvrtc_block_optimizer module~dtfft_pencil->module~dtfft_interface_cuda_runtime module~dtfft_plan dtfft_plan module~dtfft_plan->module~dtfft_interface_cuda_runtime module~dtfft_plan->module~dtfft_abstract_transpose_plan module~dtfft_plan->module~dtfft_config module~dtfft_plan->module~dtfft_executor_cufft_m module~dtfft_plan->module~dtfft_pencil module~dtfft_transpose_plan_cuda dtfft_transpose_plan_cuda module~dtfft_plan->module~dtfft_transpose_plan_cuda module~dtfft_plan->module~dtfft_abstract_executor module~dtfft_executor_fftw_m dtfft_executor_fftw_m module~dtfft_plan->module~dtfft_executor_fftw_m module~dtfft_executor_vkfft_m dtfft_executor_vkfft_m module~dtfft_plan->module~dtfft_executor_vkfft_m module~dtfft_transpose_plan_host dtfft_transpose_plan_host module~dtfft_plan->module~dtfft_transpose_plan_host module~dtfft_executor_mkl_m dtfft_executor_mkl_m module~dtfft_plan->module~dtfft_executor_mkl_m module~dtfft_transpose_handle_cuda dtfft_transpose_handle_cuda module~dtfft_transpose_handle_cuda->module~dtfft_interface_cuda_runtime module~dtfft_transpose_handle_cuda->module~dtfft_abstract_backend module~dtfft_transpose_handle_cuda->module~dtfft_backend_cufftmp_m module~dtfft_transpose_handle_cuda->module~dtfft_backend_mpi module~dtfft_transpose_handle_cuda->module~dtfft_backend_nccl_m module~dtfft_transpose_handle_cuda->module~dtfft_nvrtc_kernel module~dtfft_transpose_handle_cuda->module~dtfft_pencil module~dtfft_transpose_plan_cuda->module~dtfft_interface_cuda_runtime module~dtfft_transpose_plan_cuda->module~dtfft_abstract_backend module~dtfft_transpose_plan_cuda->module~dtfft_abstract_transpose_plan module~dtfft_transpose_plan_cuda->module~dtfft_config module~dtfft_transpose_plan_cuda->module~dtfft_nvrtc_kernel_cache module~dtfft_transpose_plan_cuda->module~dtfft_pencil module~dtfft_transpose_plan_cuda->module~dtfft_transpose_handle_cuda module~dtfft dtfft module~dtfft->module~dtfft_config module~dtfft->module~dtfft_pencil module~dtfft->module~dtfft_plan module~dtfft_abstract_executor->module~dtfft_pencil module~dtfft_api dtfft_api module~dtfft_api->module~dtfft_config module~dtfft_api->module~dtfft_pencil module~dtfft_api->module~dtfft_plan module~dtfft_executor_fftw_m->module~dtfft_pencil module~dtfft_executor_fftw_m->module~dtfft_abstract_executor module~dtfft_executor_vkfft_m->module~dtfft_config module~dtfft_executor_vkfft_m->module~dtfft_abstract_executor module~dtfft_transpose_handle_host dtfft_transpose_handle_host module~dtfft_transpose_handle_host->module~dtfft_pencil module~dtfft_transpose_plan_host->module~dtfft_abstract_transpose_plan module~dtfft_transpose_plan_host->module~dtfft_config module~dtfft_transpose_plan_host->module~dtfft_pencil module~dtfft_transpose_plan_host->module~dtfft_transpose_handle_host module~dtfft_executor_mkl_m->module~dtfft_abstract_executor

Variables

Type Visibility Attributes Name Initial
integer(kind=c_int), public, parameter :: cudaEventDisableTiming = 2

Enumerations

enum, bind(c)

Enumerators

enumerator:: cudaSuccess = 0
enumerator:: cudaErrorInvalidValue = 1
enumerator:: cudaErrorMemoryAllocation = 2
enumerator:: cudaErrorInitializationError = 3
enumerator:: cudaErrorCudartUnloading = 4
enumerator:: cudaErrorProfilerDisabled = 5
enumerator:: cudaErrorProfilerNotInitialized = 6
enumerator:: cudaErrorProfilerAlreadyStarted = 7
enumerator:: cudaErrorProfilerAlreadyStopped = 8
enumerator:: cudaErrorInvalidConfiguration = 9
enumerator:: cudaErrorInvalidPitchValue = 12
enumerator:: cudaErrorInvalidSymbol = 13
enumerator:: cudaErrorInvalidHostPointer = 16
enumerator:: cudaErrorInvalidDevicePointer = 17
enumerator:: cudaErrorInvalidTexture = 18
enumerator:: cudaErrorInvalidTextureBinding = 19
enumerator:: cudaErrorInvalidChannelDescriptor = 20
enumerator:: cudaErrorInvalidMemcpyDirection = 21
enumerator:: cudaErrorAddressOfConstant = 22
enumerator:: cudaErrorTextureFetchFailed = 23
enumerator:: cudaErrorTextureNotBound = 24
enumerator:: cudaErrorSynchronizationError = 25
enumerator:: cudaErrorInvalidFilterSetting = 26
enumerator:: cudaErrorInvalidNormSetting = 27
enumerator:: cudaErrorMixedDeviceExecution = 28
enumerator:: cudaErrorNotYetImplemented = 31
enumerator:: cudaErrorMemoryValueTooLarge = 32
enumerator:: cudaErrorInsufficientDriver = 35
enumerator:: cudaErrorInvalidSurface = 37
enumerator:: cudaErrorDuplicateVariableName = 43
enumerator:: cudaErrorDuplicateTextureName = 44
enumerator:: cudaErrorDuplicateSurfaceName = 45
enumerator:: cudaErrorDevicesUnavailable = 46
enumerator:: cudaErrorIncompatibleDriverContext = 49
enumerator:: cudaErrorMissingConfiguration = 52
enumerator:: cudaErrorPriorLaunchFailure = 53
enumerator:: cudaErrorLaunchMaxDepthExceeded = 65
enumerator:: cudaErrorLaunchFileScopedTex = 66
enumerator:: cudaErrorLaunchFileScopedSurf = 67
enumerator:: cudaErrorSyncDepthExceeded = 68
enumerator:: cudaErrorLaunchPendingCountExceeded = 69
enumerator:: cudaErrorInvalidDeviceFunction = 98
enumerator:: cudaErrorNoDevice = 100
enumerator:: cudaErrorInvalidDevice = 101
enumerator:: cudaErrorStartupFailure = 127
enumerator:: cudaErrorInvalidKernelImage = 200
enumerator:: cudaErrorDeviceUninitialized = 201
enumerator:: cudaErrorMapBufferObjectFailed = 205
enumerator:: cudaErrorUnmapBufferObjectFailed = 206
enumerator:: cudaErrorArrayIsMapped = 207
enumerator:: cudaErrorAlreadyMapped = 208
enumerator:: cudaErrorNoKernelImageForDevice = 209
enumerator:: cudaErrorAlreadyAcquired = 210
enumerator:: cudaErrorNotMapped = 211
enumerator:: cudaErrorNotMappedAsArray = 212
enumerator:: cudaErrorNotMappedAsPointer = 213
enumerator:: cudaErrorECCUncorrectable = 214
enumerator:: cudaErrorUnsupportedLimit = 215
enumerator:: cudaErrorDeviceAlreadyInUse = 216
enumerator:: cudaErrorPeerAccessUnsupported = 217
enumerator:: cudaErrorInvalidPtx = 218
enumerator:: cudaErrorInvalidGraphicsContext = 219
enumerator:: cudaErrorNvlinkUncorrectable = 220
enumerator:: cudaErrorJitCompilerNotFound = 221
enumerator:: cudaErrorInvalidSource = 300
enumerator:: cudaErrorFileNotFound = 301
enumerator:: cudaErrorSharedObjectSymbolNotFound = 302
enumerator:: cudaErrorSharedObjectInitFailed = 303
enumerator:: cudaErrorOperatingSystem = 304
enumerator:: cudaErrorInvalidResourceHandle = 400
enumerator:: cudaErrorIllegalState = 401
enumerator:: cudaErrorSymbolNotFound = 500
enumerator:: cudaErrorNotReady = 600
enumerator:: cudaErrorIllegalAddress = 700
enumerator:: cudaErrorLaunchOutOfResources = 701
enumerator:: cudaErrorLaunchTimeout = 702
enumerator:: cudaErrorLaunchIncompatibleTexturing = 703
enumerator:: cudaErrorPeerAccessAlreadyEnabled = 704
enumerator:: cudaErrorPeerAccessNotEnabled = 705
enumerator:: cudaErrorSetOnActiveProcess = 708
enumerator:: cudaErrorContextIsDestroyed = 709
enumerator:: cudaErrorAssert = 710
enumerator:: cudaErrorTooManyPeers = 711
enumerator:: cudaErrorHostMemoryAlreadyRegistered = 712
enumerator:: cudaErrorHostMemoryNotRegistered = 713
enumerator:: cudaErrorHardwareStackError = 714
enumerator:: cudaErrorIllegalInstruction = 715
enumerator:: cudaErrorMisalignedAddress = 716
enumerator:: cudaErrorInvalidAddressSpace = 717
enumerator:: cudaErrorInvalidPc = 718
enumerator:: cudaErrorLaunchFailure = 719
enumerator:: cudaErrorCooperativeLaunchTooLarge = 720
enumerator:: cudaErrorNotPermitted = 800
enumerator:: cudaErrorNotSupported = 801
enumerator:: cudaErrorSystemNotReady = 802
enumerator:: cudaErrorSystemDriverMismatch = 803
enumerator:: cudaErrorCompatNotSupportedOnDevice = 804
enumerator:: cudaErrorStreamCaptureUnsupported = 900
enumerator:: cudaErrorStreamCaptureInvalidated = 901
enumerator:: cudaErrorStreamCaptureMerge = 902
enumerator:: cudaErrorStreamCaptureUnmatched = 903
enumerator:: cudaErrorStreamCaptureUnjoined = 904
enumerator:: cudaErrorStreamCaptureIsolation = 905
enumerator:: cudaErrorStreamCaptureImplicit = 906
enumerator:: cudaErrorCapturedEvent = 907
enumerator:: cudaErrorStreamCaptureWrongThread = 908
enumerator:: cudaErrorTimeout = 909
enumerator:: cudaErrorGraphExecUpdateFailure = 910
enumerator:: cudaErrorUnknown = 999
enumerator:: cudaErrorApiFailureBase = 10000

enum, bind(c)

Enumerators

enumerator:: cudaMemcpyHostToHost = 0
enumerator:: cudaMemcpyHostToDevice = 1
enumerator:: cudaMemcpyDeviceToHost = 2
enumerator:: cudaMemcpyDeviceToDevice = 3
enumerator:: cudaMemcpyDefault = 4

Interfaces

interface

  • public function cudaStreamQuery(stream) result(cudaError_t) bind(C, name="cudaStreamQuery")

    Queries an asynchronous stream for completion status.

    Arguments

    Type IntentOptional Attributes Name
    type(dtfft_stream_t), value :: stream

    Stream identifier

    Return Value integer(kind=c_int)

    Returns cudaSuccess if all operations in stream have completed, or cudaErrorNotReady if not.

interface

  • public function cudaStreamCreate(stream) result(cudaError_t) bind(C, name="cudaStreamCreate")

    Creates an asynchronous stream.

    Arguments

    Type IntentOptional Attributes Name
    type(dtfft_stream_t) :: stream

    Pointer to the created stream

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the stream was created successfully, or an error code if there was an issue.

interface

  • public function cudaStreamDestroy(stream) result(cudaError_t) bind(C, name="cudaStreamDestroy")

    Destroys an asynchronous stream.

    Arguments

    Type IntentOptional Attributes Name
    type(dtfft_stream_t), value :: stream

    Stream identifier

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the stream was destroyed successfully, or an error code if there was an issue.

interface

  • public function cudaStreamSynchronize(stream) result(cudaError_t) bind(C, name="cudaStreamSynchronize")

    Waits for stream tasks to complete.

    Arguments

    Type IntentOptional Attributes Name
    type(dtfft_stream_t), value :: stream

    Stream identifier

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the stream tasks completed successfully, or an error code if there was an issue.

interface

  • public function cudaMalloc(ptr, count) result(cudaError_t) bind(C, name="cudaMalloc")

    Allocates memory on the device.

    Arguments

    Type IntentOptional Attributes Name
    type(c_ptr) :: ptr

    Pointer to allocated device memory
    integer(kind=c_size_t), value :: count

    Requested allocation size in bytes

    Return Value integer(kind=c_int)

    Returns cudaSuccess if memory was allocated successfully, or cudaErrorMemoryAllocation if the memory could not be allocated.

interface

  • public function cudaFree(ptr) result(cudaError_t) bind(C, name="cudaFree")

    Frees memory on the device.

    Arguments

    Type IntentOptional Attributes Name
    type(c_ptr), value :: ptr

    Pointer to memory to free

    Return Value integer(kind=c_int)

    Returns cudaSuccess if memory was freed successfully, or an error code if there was an issue.

interface

  • public function cudaMemset(ptr, val, count) result(cudaError_t) bind(C, name="cudaMemset")

    Initializes or sets device memory to a value.

    Arguments

    Type IntentOptional Attributes Name
    type(c_ptr), value :: ptr

    Pointer to device memory
    integer(kind=c_int), value :: val

    Value to set
    integer(kind=c_size_t), value :: count

    Size in bytes to set

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the memory was set successfully, or an error code if there was an issue.

interface

  • public function cudaEventCreateWithFlags(event, flags) result(cudaError_t) bind(C, name="cudaEventCreateWithFlags")

    Creates an event with the specified flags.

    Arguments

    Type IntentOptional Attributes Name
    type(cudaEvent) :: event

    Event identifier
    integer(kind=c_int), value :: flags

    Flags for event creation

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the event was created successfully, or an error code if there was an issue.

interface

  • public function cudaEventRecord(event, stream) result(cudaError_t) bind(C, name="cudaEventRecord")

    Records an event in a stream.

    Arguments

    Type IntentOptional Attributes Name
    type(cudaEvent), value :: event

    Event identifier
    type(dtfft_stream_t), value :: stream

    Stream identifier

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the event was recorded successfully, or an error code if there was an issue.

interface

  • public function cudaStreamWaitEvent(stream, event, flags) result(cudaError_t) bind(C, name="cudaStreamWaitEvent")

    Makes a stream wait on an event.

    Arguments

    Type IntentOptional Attributes Name
    type(dtfft_stream_t), value :: stream

    Stream identifier
    type(cudaEvent), value :: event

    Event identifier
    integer(kind=c_int), value :: flags

    Flags for the wait operation

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the stream is waiting successfully, or an error code if there was an issue.

interface

  • public function cudaEventDestroy(event) result(cudaError_t) bind(C, name="cudaEventDestroy")

    Destroys an event.

    Arguments

    Type IntentOptional Attributes Name
    type(cudaEvent), value :: event

    Event identifier

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the event was destroyed successfully, or an error code if there was an issue.

interface

  • public function cudaEventCreate(event) result(cudaError_t) bind(C, name="cudaEventCreate")

    Creates an event.

    Arguments

    Type IntentOptional Attributes Name
    type(cudaEvent) :: event

    Event identifier

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the event was created successfully, or an error code if there was an issue.

interface

  • public function cudaEventSynchronize(event) result(cudaError_t) bind(C, name="cudaEventSynchronize")

    Waits for an event to complete.

    Arguments

    Type IntentOptional Attributes Name
    type(cudaEvent), value :: event

    Event identifier

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the event completed successfully, or an error code if there was an issue.

interface

  • public function cudaEventElapsedTime(time, start, end) result(cudaError_t) bind(C, name="cudaEventElapsedTime")

    Computes the elapsed time between two events.

    Arguments

    Type IntentOptional Attributes Name
    real(kind=c_float) :: time

    Elapsed time in milliseconds
    type(cudaEvent), value :: start

    Starting event
    type(cudaEvent), value :: end

    Ending event

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the elapsed time was computed successfully, or an error code if there was an issue.

public interface cudaMemcpyAsync

Copies data asynchronously between host and device.

  • private function cudaMemcpyAsync_ptr(dst, src, count, kdir, stream) result(cudaError_t) bind(C, name="cudaMemcpyAsync")

    Arguments

    Type IntentOptional Attributes Name
    type(c_ptr), value :: dst

    Destination pointer
    type(c_ptr), value :: src

    Source pointer
    integer(kind=c_size_t), value :: count

    Size in bytes to copy
    integer(kind=c_int), value :: kdir

    Direction of copy (host-to-device, device-to-host, etc.)
    type(dtfft_stream_t), value :: stream

    Stream identifier

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the copy was initiated successfully, or an error code if there was an issue.

  • private function cudaMemcpyAsync_r32(dst, src, count, kdir, stream) result(cudaError_t) bind(C, name="cudaMemcpyAsync")

    Arguments

    Type IntentOptional Attributes Name
    real(kind=c_float) :: dst

    Destination array (32-bit float)
    real(kind=c_float) :: src

    Source array (32-bit float)
    integer(kind=c_size_t), value :: count

    Number of elements to copy
    integer(kind=c_int), value :: kdir

    Direction of copy
    type(dtfft_stream_t), value :: stream

    Stream identifier

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the copy was initiated successfully, or an error code if there was an issue.

public interface cudaMemcpy

Copies data synchronously between host and device.

  • private function cudaMemcpy_ptr(dst, src, count, kdir) result(cudaError_t) bind(C, name="cudaMemcpy")

    Arguments

    Type IntentOptional Attributes Name
    type(c_ptr), value :: dst

    Destination pointer
    type(c_ptr), value :: src

    Source pointer
    integer(kind=c_size_t), value :: count

    Size in bytes to copy
    integer(kind=c_int), value :: kdir

    Direction of copy

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the copy was completed successfully, or an error code if there was an issue.

  • private function cudaMemcpy_r64(dst, src, count, kdir) result(cudaError_t) bind(C, name="cudaMemcpy")

    Arguments

    Type IntentOptional Attributes Name
    real(kind=c_double) :: dst(*)

    Destination array (64-bit float)
    real(kind=c_double) :: src(*)

    Source array (64-bit float)
    integer(kind=c_size_t), value :: count

    Number of bytes to copy
    integer(kind=c_int), value :: kdir

    Direction of copy

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the copy was completed successfully, or an error code if there was an issue.

  • private function cudaMemcpy_r32(dst, src, count, kdir) result(cudaError_t) bind(C, name="cudaMemcpy")

    Arguments

    Type IntentOptional Attributes Name
    real(kind=c_float) :: dst(*)

    Destination array (32-bit float)
    real(kind=c_float) :: src(*)

    Source array (32-bit float)
    integer(kind=c_size_t), value :: count

    Number of bytes to copy
    integer(kind=c_int), value :: kdir

    Direction of copy

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the copy was completed successfully, or an error code if there was an issue.

interface

  • public function cudaGetDevice(num) result(cudaError_t) bind(C, name="cudaGetDevice")

    Returns the current device.

    Arguments

    Type IntentOptional Attributes Name
    integer(kind=c_int) :: num

    Device number

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the device was retrieved successfully, or an error code if there was an issue.

interface

  • public function cudaGetDeviceCount(num) result(cudaError_t) bind(C, name="cudaGetDeviceCount")

    Returns the number of available devices.

    Arguments

    Type IntentOptional Attributes Name
    integer(kind=c_int) :: num

    Number of devices

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the device count was retrieved successfully, or an error code if there was an issue.

interface

  • public function cudaSetDevice(num) result(cudaError_t) bind(C, name="cudaSetDevice")

    Sets the current device.

    Arguments

    Type IntentOptional Attributes Name
    integer(kind=c_int), value :: num

    Device number

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the device was set successfully, or an error code if there was an issue.

interface

  • public function cudaMemGetInfo(free, total) result(cudaError_t) bind(C, name="cudaMemGetInfo")

    Returns the amount of free and total memory on the device.

    Arguments

    Type IntentOptional Attributes Name
    integer(kind=c_size_t) :: free

    Free memory in bytes
    integer(kind=c_size_t) :: total

    Total memory in bytes

    Return Value integer(kind=c_int)

    Returns cudaSuccess if the memory information was retrieved successfully, or an error code if there was an issue.

interface

  • public function cudaDeviceSynchronize() result(cudaError_t) bind(C, name="cudaDeviceSynchronize")

    Synchronizes the device, blocking until all preceding tasks in all streams have completed.

    Arguments

    None

    Return Value integer(kind=c_int)

    Returns cudaSuccess if syncronization was or an error code if there was an issue.

interface

  • public function cudaGetLastError() result(cudaError_t) bind(C, name="cudaGetLastError")

    Returns the last error from a runtime call.

    Arguments

    None

    Return Value integer(kind=c_int)

    Returns cudaSuccess if no error was detected or an error code if there was an issue.

interface

  • public subroutine get_device_props(device, props) bind(C, name="get_device_props_cuda")

    Returns the CUDA device properties for a given device number.

    Arguments

    Type IntentOptional Attributes Name
    integer(kind=c_int), value :: device

    Device number
    type(device_props) :: props

    GPU Properties

interface

  • private function cudaGetErrorString_c(errcode) result(string) bind(C, name="cudaGetErrorString")

    Returns the string representation of an error code.

    Arguments

    Type IntentOptional Attributes Name
    integer(kind=c_int), value :: errcode

    Error code

    Return Value type(c_ptr)

    Pointer to the error string


Derived Types

type, public, bind(C) ::  cudaEvent

CUDA event types

Components

Type Visibility Attributes Name Initial
type(c_ptr), public :: event

Handle

type, public, bind(C) ::  device_props

GPU device properties obtained from cudaDeviceProp

Components

Type Visibility Attributes Name Initial
integer(kind=c_int), public :: sm_count

Number of multiprocessors on device (cudaDeviceProp.multiProcessorCount)
integer(kind=c_int), public :: max_threads_per_sm

Maximum resident threads per multiprocessor (cudaDeviceProp.maxThreadsPerMultiProcessor)
integer(kind=c_int), public :: max_blocks_per_sm

Maximum number of resident blocks per multiprocessor (cudaDeviceProp.maxBlocksPerMultiProcessor)
integer(kind=c_size_t), public :: shared_mem_per_sm

Shared memory per multiprocessor (cudaDeviceProp.sharedMemPerMultiprocessor)
integer(kind=c_int), public :: max_threads_per_block

Maximum number of threads per block (cudaDeviceProp.maxThreadsPerBlock)
integer(kind=c_size_t), public :: shared_mem_per_block

Shared memory available per block in bytes (cudaDeviceProp.sharedMemPerBlock)
integer(kind=c_int), public :: l2_cache_size

Size of L2 cache in bytes (cudaDeviceProp.l2CacheSize)
integer(kind=c_int), public :: compute_capability_major

Major compute capability (cudaDeviceProp.major)
integer(kind=c_int), public :: compute_capability_minor

Minor compute capability (cudaDeviceProp.minor)

Functions

public function cudaGetErrorString(errcode) result(string)

Helper function that returns a string describing the given nvrtcResult code If the error code is not recognized, “unrecognized error code” is returned.

Arguments

Type IntentOptional Attributes Name
integer(kind=c_int), intent(in) :: errcode

CUDA Runtime Compilation API result code.

Return Value character(len=:), allocatable

Result string