Documentation
¶
Index ¶
- Constants
- func DeviceComputeCapability(device Device) (major, minor int)
- func DeviceGetAttribute(attrib DeviceAttribute, dev Device) int
- func DeviceGetCount() int
- func DeviceGetName(dev Device) string
- func DeviceReset()
- func DeviceTotalMem(device Device) int64
- func FreeHost(ptr unsafe.Pointer)
- func Init(flags int)
- func MallocHost(bytes int64) unsafe.Pointer
- func MemCpy(dst, src unsafe.Pointer, bytes int64, flags uint)
- func SetDevice(device Device)
- func SetDeviceFlags(flags uint)
- func Version() int
- type DevProp
- type Device
- type DeviceAttribute
- type DevicePtr
- type Result
Constants ¶
View Source
const ( // The default, decides to yield or not based on active CUDA threads and processors. DeviceAuto = C.cudaDeviceScheduleAuto // Actively spin while waiting for device. DeviceSpin = C.cudaDeviceScheduleSpin // Yield when waiting. DeviceYield = C.cudaDeviceScheduleYield // ScheduleBlockingSync block CPU on sync. DeviceScheduleBlockingSync = C.cudaDeviceScheduleBlockingSync // ScheduleBlockingSync block CPU on sync. Deprecated since cuda 4.0 DeviceBlockingSync = C.cudaDeviceBlockingSync // For use with pinned host memory DeviceMapHost = C.cudaDeviceMapHost // Do not reduce local memory to try and prevent thrashing DeviceLmemResizeToMax = C.cudaDeviceLmemResizeToMax )
Flags for SetDeviceFlags
View Source
const ( // Host to Host HtoH = C.cudaMemcpyHostToHost // Host to Device HtoD = C.cudaMemcpyHostToDevice // Device to Host DtoH = C.cudaMemcpyDeviceToHost // Device to Device DtoD = C.cudaMemcpyDeviceToDevice // Default, unified virtual address space Virt = C.cudaMemcpyDefault )
Flags for memory copy types
Variables ¶
This section is empty.
Functions ¶
func DeviceComputeCapability ¶
Returns the compute capability of the device.
func DeviceGetAttribute ¶
func DeviceGetAttribute(attrib DeviceAttribute, dev Device) int
Gets the value of a device attribute.
func DeviceGetCount ¶
func DeviceGetCount() int
Returns the number of devices with compute capability greater than or equal to 1.0 that are available for execution.
func DeviceTotalMem ¶
Returns the total amount of memory available on the device in bytes.
func Init ¶
func Init(flags int)
Initialize the CUDA driver API. Currently, flags must be 0. If Init() has not been called, any function from the driver API will panic with ERROR_NOT_INITIALIZED.
func MallocHost ¶
Types ¶
type DevProp ¶
type DevProp struct {
MaxThreadsPerBlock int
MaxThreadsDim [3]int
MaxGridSize [3]int
TotalConstantMemory int
SIMDWidth int
MemPitch int
RegsPerBlock int
ClockRate int
TextureAlign int
}
Device properties
func DeviceGetProperties ¶
Returns the device's properties.
type Device ¶
type Device int
CUDA Device number.
func (Device) Attribute ¶
func (dev Device) Attribute(attrib DeviceAttribute) int
Gets the value of a device attribute.
func (Device) ComputeCapability ¶
Returns the compute capability of the device.
type DeviceAttribute ¶
type DeviceAttribute int
const ( MAX_THREADS_PER_BLOCK DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAX_THREADS_PER_BLOCK // Maximum number of threads per block MAX_BLOCK_DIM_X DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAX_BLOCK_DIM_X // Maximum block dimension X MAX_BLOCK_DIM_Y DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAX_BLOCK_DIM_Y // Maximum block dimension Y MAX_BLOCK_DIM_Z DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAX_BLOCK_DIM_Z // Maximum block dimension Z MAX_GRID_DIM_X DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAX_GRID_DIM_X // Maximum grid dimension X MAX_GRID_DIM_Y DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAX_GRID_DIM_Y // Maximum grid dimension Y MAX_GRID_DIM_Z DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAX_GRID_DIM_Z // Maximum grid dimension Z MAX_SHARED_MEMORY_PER_BLOCK DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAX_SHARED_MEMORY_PER_BLOCK // Maximum shared memory available per block in bytes TOTAL_CONSTANT_MEMORY DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_TOTAL_CONSTANT_MEMORY // Memory available on device for __constant__ variables in a CUDA C kernel in bytes WARP_SIZE DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_WARP_SIZE // Warp size in threads MAX_PITCH DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAX_PITCH // Maximum pitch in bytes allowed by memory copies MAX_REGISTERS_PER_BLOCK DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAX_REGISTERS_PER_BLOCK // Maximum number of 32-bit registers available per block CLOCK_RATE DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_CLOCK_RATE // Peak clock frequency in kilohertz TEXTURE_ALIGNMENT DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_TEXTURE_ALIGNMENT // Alignment requirement for textures MULTIPROCESSOR_COUNT DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MULTIPROCESSOR_COUNT // Number of multiprocessors on device KERNEL_EXEC_TIMEOUT DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_KERNEL_EXEC_TIMEOUT // Specifies whether there is a run time limit on kernels INTEGRATED DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_INTEGRATED // Device is integrated with host memory CAN_MAP_HOST_MEMORY DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_CAN_MAP_HOST_MEMORY // Device can map host memory into CUDA address space COMPUTE_MODE DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_COMPUTE_MODE // Compute mode (See ::CUcomputemode for details) MAXIMUM_TEXTURE1D_WIDTH DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAXIMUM_TEXTURE1D_WIDTH // Maximum 1D texture width MAXIMUM_TEXTURE2D_WIDTH DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAXIMUM_TEXTURE2D_WIDTH // Maximum 2D texture width MAXIMUM_TEXTURE2D_HEIGHT DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAXIMUM_TEXTURE2D_HEIGHT // Maximum 2D texture height MAXIMUM_TEXTURE3D_WIDTH DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAXIMUM_TEXTURE3D_WIDTH // Maximum 3D texture width MAXIMUM_TEXTURE3D_HEIGHT DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAXIMUM_TEXTURE3D_HEIGHT // Maximum 3D texture height MAXIMUM_TEXTURE3D_DEPTH DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAXIMUM_TEXTURE3D_DEPTH // Maximum 3D texture depth MAXIMUM_TEXTURE2D_LAYERED_WIDTH DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAXIMUM_TEXTURE2D_LAYERED_WIDTH // Maximum 2D layered texture width MAXIMUM_TEXTURE2D_LAYERED_HEIGHT DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAXIMUM_TEXTURE2D_LAYERED_HEIGHT // Maximum 2D layered texture height MAXIMUM_TEXTURE2D_LAYERED_LAYERS DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAXIMUM_TEXTURE2D_LAYERED_LAYERS // Maximum layers in a 2D layered texture SURFACE_ALIGNMENT DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_SURFACE_ALIGNMENT // Alignment requirement for surfaces CONCURRENT_KERNELS DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_CONCURRENT_KERNELS // Device can possibly execute multiple kernels concurrently ECC_ENABLED DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_ECC_ENABLED // Device has ECC support enabled PCI_BUS_ID DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_PCI_BUS_ID // PCI bus ID of the device PCI_DEVICE_ID DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_PCI_DEVICE_ID // PCI device ID of the device TCC_DRIVER DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_TCC_DRIVER // Device is using TCC driver model MEMORY_CLOCK_RATE DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MEMORY_CLOCK_RATE // Peak memory clock frequency in kilohertz GLOBAL_MEMORY_BUS_WIDTH DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_GLOBAL_MEMORY_BUS_WIDTH // Global memory bus width in bits L2_CACHE_SIZE DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_L2_CACHE_SIZE // Size of L2 cache in bytes MAX_THREADS_PER_MULTIPROCESSOR DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAX_THREADS_PER_MULTIPROCESSOR // Maximum resident threads per multiprocessor ASYNC_ENGINE_COUNT DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_ASYNC_ENGINE_COUNT // Number of asynchronous engines UNIFIED_ADDRESSING DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_UNIFIED_ADDRESSING // Device uses shares a unified address space with the host MAXIMUM_TEXTURE1D_LAYERED_WIDTH DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAXIMUM_TEXTURE1D_LAYERED_WIDTH // Maximum 1D layered texture width MAXIMUM_TEXTURE1D_LAYERED_LAYERS DeviceAttribute = C.CU_DEVICE_ATTRIBUTE_MAXIMUM_TEXTURE1D_LAYERED_LAYERS // Maximum layers in a 1D layered texture )
type Result ¶
type Result int
CUDA error status. CUDA error statuses are not returned by functions but checked and passed to panic() when not successful. If desired, they can be caught by recover().
const ( SUCCESS Result = C.CUDA_SUCCESS ERROR_INVALID_VALUE Result = C.CUDA_ERROR_INVALID_VALUE ERROR_OUT_OF_MEMORY Result = C.CUDA_ERROR_OUT_OF_MEMORY ERROR_NOT_INITIALIZED Result = C.CUDA_ERROR_NOT_INITIALIZED ERROR_DEINITIALIZED Result = C.CUDA_ERROR_DEINITIALIZED ERROR_PROFILER_DISABLED Result = C.CUDA_ERROR_PROFILER_DISABLED ERROR_PROFILER_NOT_INITIALIZED Result = C.CUDA_ERROR_PROFILER_NOT_INITIALIZED ERROR_PROFILER_ALREADY_STARTED Result = C.CUDA_ERROR_PROFILER_ALREADY_STARTED ERROR_PROFILER_ALREADY_STOPPED Result = C.CUDA_ERROR_PROFILER_ALREADY_STOPPED ERROR_NO_DEVICE Result = C.CUDA_ERROR_NO_DEVICE ERROR_INVALID_DEVICE Result = C.CUDA_ERROR_INVALID_DEVICE ERROR_INVALID_IMAGE Result = C.CUDA_ERROR_INVALID_IMAGE ERROR_INVALID_CONTEXT Result = C.CUDA_ERROR_INVALID_CONTEXT ERROR_CONTEXT_ALREADY_CURRENT Result = C.CUDA_ERROR_CONTEXT_ALREADY_CURRENT ERROR_MAP_FAILED Result = C.CUDA_ERROR_MAP_FAILED ERROR_UNMAP_FAILED Result = C.CUDA_ERROR_UNMAP_FAILED ERROR_ARRAY_IS_MAPPED Result = C.CUDA_ERROR_ARRAY_IS_MAPPED ERROR_ALREADY_MAPPED Result = C.CUDA_ERROR_ALREADY_MAPPED ERROR_NO_BINARY_FOR_GPU Result = C.CUDA_ERROR_NO_BINARY_FOR_GPU ERROR_ALREADY_ACQUIRED Result = C.CUDA_ERROR_ALREADY_ACQUIRED ERROR_NOT_MAPPED Result = C.CUDA_ERROR_NOT_MAPPED ERROR_NOT_MAPPED_AS_ARRAY Result = C.CUDA_ERROR_NOT_MAPPED_AS_ARRAY ERROR_NOT_MAPPED_AS_POINTER Result = C.CUDA_ERROR_NOT_MAPPED_AS_POINTER ERROR_ECC_UNCORRECTABLE Result = C.CUDA_ERROR_ECC_UNCORRECTABLE ERROR_UNSUPPORTED_LIMIT Result = C.CUDA_ERROR_UNSUPPORTED_LIMIT ERROR_CONTEXT_ALREADY_IN_USE Result = C.CUDA_ERROR_CONTEXT_ALREADY_IN_USE ERROR_INVALID_SOURCE Result = C.CUDA_ERROR_INVALID_SOURCE ERROR_FILE_NOT_FOUND Result = C.CUDA_ERROR_FILE_NOT_FOUND ERROR_SHARED_OBJECT_SYMBOL_NOT_FOUND Result = C.CUDA_ERROR_SHARED_OBJECT_SYMBOL_NOT_FOUND ERROR_SHARED_OBJECT_INIT_FAILED Result = C.CUDA_ERROR_SHARED_OBJECT_INIT_FAILED ERROR_OPERATING_SYSTEM Result = C.CUDA_ERROR_OPERATING_SYSTEM ERROR_INVALID_HANDLE Result = C.CUDA_ERROR_INVALID_HANDLE ERROR_NOT_FOUND Result = C.CUDA_ERROR_NOT_FOUND ERROR_NOT_READY Result = C.CUDA_ERROR_NOT_READY ERROR_LAUNCH_FAILED Result = C.CUDA_ERROR_LAUNCH_FAILED ERROR_LAUNCH_OUT_OF_RESOURCES Result = C.CUDA_ERROR_LAUNCH_OUT_OF_RESOURCES ERROR_LAUNCH_TIMEOUT Result = C.CUDA_ERROR_LAUNCH_TIMEOUT ERROR_LAUNCH_INCOMPATIBLE_TEXTURING Result = C.CUDA_ERROR_LAUNCH_INCOMPATIBLE_TEXTURING ERROR_PEER_ACCESS_ALREADY_ENABLED Result = C.CUDA_ERROR_PEER_ACCESS_ALREADY_ENABLED ERROR_PEER_ACCESS_NOT_ENABLED Result = C.CUDA_ERROR_PEER_ACCESS_NOT_ENABLED ERROR_PRIMARY_CONTEXT_ACTIVE Result = C.CUDA_ERROR_PRIMARY_CONTEXT_ACTIVE ERROR_CONTEXT_IS_DESTROYED Result = C.CUDA_ERROR_CONTEXT_IS_DESTROYED ERROR_ASSERT Result = C.CUDA_ERROR_ASSERT ERROR_TOO_MANY_PEERS Result = C.CUDA_ERROR_TOO_MANY_PEERS ERROR_HOST_MEMORY_ALREADY_REGISTERED Result = C.CUDA_ERROR_HOST_MEMORY_ALREADY_REGISTERED ERROR_HOST_MEMORY_NOT_REGISTERED Result = C.CUDA_ERROR_HOST_MEMORY_NOT_REGISTERED ERROR_HARDWARE_STACK_ERROR Result = 714 //C.CUDA_ERROR_HARDWARE_STACK_ERROR ERROR_ILLEGAL_INSTRUCTION Result = 715 //C.CUDA_ERROR_ILLEGAL_INSTRUCTION ERROR_MISALIGNED_ADDRESS Result = 716 //C.CUDA_ERROR_MISALIGNED_ADDRESS ERROR_INVALID_ADDRESS_SPACE Result = 717 //C.CUDA_ERROR_INVALID_ADDRESS_SPACE ERROR_INVALID_PC Result = 718 //C.CUDA_ERROR_INVALID_PC ERROR_NOT_PERMITTED Result = 800 //C.CUDA_ERROR_NOT_PERMITTED ERROR_NOT_SUPPORTED Result = 801 //C.CUDA_ERROR_NOT_SUPPORTED ERROR_UNKNOWN Result = C.CUDA_ERROR_UNKNOWN )
Source Files
Click to show internal directories.
Click to hide internal directories.