tf_cuda.h File Reference

#include "tfError.h"
#include <cuda.h>
#include <cuda_runtime.h>
#include <nvrtc.h>
#include <vector>
#include <stdexcept>
#include <string>

Include dependency graph for tf_cuda.h:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Classes
struct	TissueForge::cuda::CUDARTSource
	Convenience class for loading source from file and storing, here intended for CUDA. More...
struct	TissueForge::cuda::CUDARTProgram
	A JIT-compiled CUDA Tissue Forge program. More...
struct	TissueForge::cuda::CUDAFunction
	A CUDA kernel from a JIT-compiled Tissue Forge program. More...
struct	TissueForge::cuda::CUDAContext
	A convenience wrap of the CUDA context for JIT-compiled Tissue Forge programs. More...
struct	TissueForge::cuda::CUDADevice
	A simple interface with a CUDA device. More...

Namespaces
namespace	TissueForge
	Include Python header, disable linking to pythonX_d.lib on Windows in debug mode.
namespace	TissueForge::cuda
	Tissue Forge GPU acceleration on CUDA-supporting devices.

Macros
#define	TF_CUDA_CALL(res)   cuda_errorchk(res, __FILE__, __LINE__)
#define	TF_NVRTC_CALL(res)   nvrtc_errorchk(res, __FILE__, __LINE__)
#define	TF_CUDART_CALL(res)   cudart_errorchk(res, __FILE__, __LINE__)

Enumerations
enum	ErrorCode : int {   TFCUDAERR_ok = 0 , TFCUDAERR_setdevice , TFCUDAERR_setblocks , TFCUDAERR_setthreads ,   TFCUDAERR_ondevice , TFCUDAERR_notondevice , TFCUDAERR_cleardevices , TFCUDAERR_refresh ,   TFCUDAERR_send , TFCUDAERR_pull , TFCUDAERR_LAST }

Functions
CUresult	TissueForge::cuda::cuda_errorchk (CUresult retCode, const char *file, int line)
nvrtcResult	TissueForge::cuda::nvrtc_errorchk (nvrtcResult retCode, const char *file, int line)
cudaError_t	TissueForge::cuda::cudart_errorchk (cudaError_t retCode, const char *file, int line)
void	TissueForge::cuda::init ()
	Initialize CUDA.
void	TissueForge::cuda::setGLDevice (const int &deviceId)
std::string	TissueForge::cuda::getDeviceName (const int &deviceId)
	Get the name of a device.
size_t	TissueForge::cuda::getDeviceTotalMem (const int &deviceId)
	Get the total memory of device.
int	TissueForge::cuda::getDeviceAttribute (const int &deviceId, const int &attrib)
	Get the attribute value of a device.
int	TissueForge::cuda::getNumDevices ()
	Get number of available compute-capable devices.
std::string	TissueForge::cuda::getDevicePCIBusId (const int &deviceId)
	Get the PCI bus id of a device.
int	TissueForge::cuda::getCurrentDevice ()
	Get the device id of the current context of the calling CPU thread.
int	TissueForge::cuda::maxThreadsPerBlock (const int &deviceId)
	Maximum number of threads per block.
int	TissueForge::cuda::maxBlockDimX (const int &deviceId)
	Maximum x-dimension of a block.
int	TissueForge::cuda::maxBlockDimY (const int &deviceId)
	Maximum y-dimension of a block.
int	TissueForge::cuda::maxBlockDimZ (const int &deviceId)
	Maximum z-dimension of a block.
int	TissueForge::cuda::maxGridDimX (const int &deviceId)
	Maximum x-dimension of a grid.
int	TissueForge::cuda::maxGridDimY (const int &deviceId)
	Maximum y-dimension of a grid.
int	TissueForge::cuda::maxGridDimZ (const int &deviceId)
	Maximum z-dimension of a grid.
int	TissueForge::cuda::maxSharedMemPerBlock (const int &deviceId)
	Maximum amount of shared memory available to a thread block in bytes.
int	TissueForge::cuda::maxTotalMemConst (const int &deviceId)
	Memory available on device for constant variables in a CUDA C kernel in bytes.
int	TissueForge::cuda::warpSize (const int &deviceId)
	Warp size in threads.
int	TissueForge::cuda::maxRegsPerBlock (const int &deviceId)
	Maximum number of 32-bit registers available to a thread block.
int	TissueForge::cuda::clockRate (const int &deviceId)
	The typical clock frequency in kilohertz.
bool	TissueForge::cuda::gpuOverlap (const int &deviceId)
	Test if the device can concurrently copy memory between host and device while executing a kernel.
int	TissueForge::cuda::numMultiprocessors (const int &deviceId)
	Number of multiprocessors on the device.
bool	TissueForge::cuda::kernelExecTimeout (const int &deviceId)
	Test if there is a run time limit for kernels executed on the device.
bool	TissueForge::cuda::computeModeDefault (const int &deviceId)
	Test if device is not restricted and can have multiple CUDA contexts present at a single time.
bool	TissueForge::cuda::computeModeProhibited (const int &deviceId)
	Test if device is prohibited from creating new CUDA contexts.
bool	TissueForge::cuda::computeModeExclusive (const int &deviceId)
	Test if device can have only one context used by a single process at a time.
int	TissueForge::cuda::PCIDeviceId (const int &deviceId)
	PCI device (also known as slot) identifier of the device.
int	TissueForge::cuda::PCIDomainId (const int &deviceId)
	PCI domain identifier of the device.
int	TissueForge::cuda::clockRateMem (const int &deviceId)
	Peak memory clock frequency in kilohertz.
int	TissueForge::cuda::globalMemBusWidth (const int &deviceId)
	Global memory bus width in bits.
int	TissueForge::cuda::L2CacheSize (const int &deviceId)
	Size of L2 cache in bytes. 0 if the device doesn't have L2 cache.
int	TissueForge::cuda::maxThreadsPerMultiprocessor (const int &deviceId)
	Maximum resident threads per multiprocessor.
int	TissueForge::cuda::computeCapabilityMajor (const int &deviceId)
	Major compute capability version number.
int	TissueForge::cuda::computeCapabilityMinor (const int &deviceId)
	Minor compute capability version number.
bool	TissueForge::cuda::L1CacheSupportGlobal (const int &deviceId)
	Test if device supports caching globals in L1 cache.
bool	TissueForge::cuda::L1CacheSupportLocal (const int &deviceId)
	Test if device supports caching locals in L1 cache.
int	TissueForge::cuda::maxSharedMemPerMultiprocessor (const int &deviceId)
	Maximum amount of shared memory available to a multiprocessor in bytes.
int	TissueForge::cuda::maxRegsPerMultiprocessor (const int &deviceId)
	Maximum number of 32-bit registers available to a multiprocessor.
bool	TissueForge::cuda::managedMem (const int &deviceId)
	Test if device supports allocating managed memory on this system.
bool	TissueForge::cuda::multiGPUBoard (const int &deviceId)
	Test if device is on a multi-GPU board.
int	TissueForge::cuda::multiGPUBoardGroupId (const int &deviceId)
	Unique identifier for a group of devices associated with the same board.
void	TissueForge::cuda::test (const int &numBlocks, const int &numThreads, const int &numEls, const int &deviceId=0)
	Tests JIT-compiled program execution and deployment.
std::string	TissueForge::cuda::tfIncludePath ()
	Returns the current path to the installed Tissue Forge include directory.
HRESULT	TissueForge::cuda::setTfIncludePath (const std::string &_path)
	Set the current path to the installed Tissue Forge include directory.
std::string	TissueForge::cuda::tfPrivateIncludePath ()
	Returns the path to the installed Tissue Forge private include directory.
std::string	TissueForge::cuda::tfResourcePath ()
	Returns the current path to the installed Tissue Forge resource directory.
HRESULT	TissueForge::cuda::setTfResourcePath (const std::string &_path)
	Set the current path to the installed Tissue Forge resource directory.
std::string	TissueForge::cuda::CUDAPath ()
	Returns the path to the installed Tissue Forge CUDA resources directory.
std::string	TissueForge::cuda::CUDAIncludePath ()
	Returns the current path to the installed CUDA include directory.
HRESULT	TissueForge::cuda::setCUDAIncludePath (const std::string &_path)
	Set the current path to the installed CUDA include directory.
std::string	TissueForge::cuda::CUDAResourcePath (const std::string &relativePath)
	Returns an absolute path to a subdirectory of the install Tissue Forge CUDA resources directory.
std::string	TissueForge::cuda::CUDAPTXObjectRelPath ()
	Returns the relative path to the installed Tissue Forge CUDA PTX object directory.
std::vector< std::string >	TissueForge::cuda::CUDAArchs ()
	Returns the supported CUDA architectures of the installation.