StarPU Internal Handbook
driver_cuda.h File Reference
#include <common/config.h>
#include <cuda.h>
#include <cuda_runtime_api.h>
#include <cublas.h>
#include <starpu.h>
#include <core/workers.h>
#include <datawizard/node_ops.h>

Go to the source code of this file.

Functions

void _starpu_cuda_init (void)
 
unsigned _starpu_get_cuda_device_count (void)
 
void _starpu_cuda_discover_devices (struct _starpu_machine_config *)
 
void _starpu_init_cuda (void)
 
void * _starpu_cuda_worker (void *)
 
cudaStream_t starpu_cuda_get_local_in_transfer_stream (void)
 
cudaStream_t starpu_cuda_get_in_transfer_stream (unsigned dst_node)
 
cudaStream_t starpu_cuda_get_local_out_transfer_stream (void)
 
cudaStream_t starpu_cuda_get_out_transfer_stream (unsigned src_node)
 
cudaStream_t starpu_cuda_get_peer_transfer_stream (unsigned src_node, unsigned dst_node)
 
unsigned _starpu_cuda_test_request_completion (struct _starpu_async_channel *async_channel)
 
void _starpu_cuda_wait_request_completion (struct _starpu_async_channel *async_channel)
 
int _starpu_cuda_copy_interface_from_cpu_to_cuda (starpu_data_handle_t handle, void *src_interface, unsigned src_node, void *dst_interface, unsigned dst_node, struct _starpu_data_request *req)
 
int _starpu_cuda_copy_interface_from_cuda_to_cuda (starpu_data_handle_t handle, void *src_interface, unsigned src_node, void *dst_interface, unsigned dst_node, struct _starpu_data_request *req)
 
int _starpu_cuda_copy_interface_from_cuda_to_cpu (starpu_data_handle_t handle, void *src_interface, unsigned src_node, void *dst_interface, unsigned dst_node, struct _starpu_data_request *req)
 
int _starpu_cuda_copy_data_from_cuda_to_cuda (uintptr_t src, size_t src_offset, unsigned src_node, uintptr_t dst, size_t dst_offset, unsigned dst_node, size_t size, struct _starpu_async_channel *async_channel)
 
int _starpu_cuda_copy_data_from_cuda_to_cpu (uintptr_t src, size_t src_offset, unsigned src_node, uintptr_t dst, size_t dst_offset, unsigned dst_node, size_t size, struct _starpu_async_channel *async_channel)
 
int _starpu_cuda_copy_data_from_cpu_to_cuda (uintptr_t src, size_t src_offset, unsigned src_node, uintptr_t dst, size_t dst_offset, unsigned dst_node, size_t size, struct _starpu_async_channel *async_channel)
 
int _starpu_cuda_copy2d_data_from_cuda_to_cuda (uintptr_t src, size_t src_offset, unsigned src_node, uintptr_t dst, size_t dst_offset, unsigned dst_node, size_t blocksize, size_t numblocks, size_t ld_src, size_t ld_dst, struct _starpu_async_channel *async_channel)
 
int _starpu_cuda_copy2d_data_from_cuda_to_cpu (uintptr_t src, size_t src_offset, unsigned src_node, uintptr_t dst, size_t dst_offset, unsigned dst_node, size_t blocksize, size_t numblocks, size_t ld_src, size_t ld_dst, struct _starpu_async_channel *async_channel)
 
int _starpu_cuda_copy2d_data_from_cpu_to_cuda (uintptr_t src, size_t src_offset, unsigned src_node, uintptr_t dst, size_t dst_offset, unsigned dst_node, size_t blocksize, size_t numblocks, size_t ld_src, size_t ld_dst, struct _starpu_async_channel *async_channel)
 
int _starpu_cuda_copy3d_data_from_cuda_to_cuda (uintptr_t src, size_t src_offset, unsigned src_node, uintptr_t dst, size_t dst_offset, unsigned dst_node, size_t blocksize, size_t numblocks_1, size_t ld1_src, size_t ld1_dst, size_t numblocks_2, size_t ld2_src, size_t ld2_dst, struct _starpu_async_channel *async_channel)
 
int _starpu_cuda_copy3d_data_from_cuda_to_cpu (uintptr_t src, size_t src_offset, unsigned src_node, uintptr_t dst, size_t dst_offset, unsigned dst_node, size_t blocksize, size_t numblocks_1, size_t ld1_src, size_t ld1_dst, size_t numblocks_2, size_t ld2_src, size_t ld2_dst, struct _starpu_async_channel *async_channel)
 
int _starpu_cuda_copy3d_data_from_cpu_to_cuda (uintptr_t src, size_t src_offset, unsigned src_node, uintptr_t dst, size_t dst_offset, unsigned dst_node, size_t blocksize, size_t numblocks_1, size_t ld1_src, size_t ld1_dst, size_t numblocks_2, size_t ld2_src, size_t ld2_dst, struct _starpu_async_channel *async_channel)
 
int _starpu_cuda_is_direct_access_supported (unsigned node, unsigned handling_node)
 
uintptr_t _starpu_cuda_malloc_on_node (unsigned dst_node, size_t size, int flags)
 
void _starpu_cuda_free_on_node (unsigned dst_node, uintptr_t addr, size_t size, int flags)
 

Variables

struct _starpu_driver_ops _starpu_driver_cuda_ops
 
struct _starpu_node_ops _starpu_driver_cuda_node_ops
 
int _starpu_cuda_bus_ids [STARPU_MAXCUDADEVS+STARPU_MAXNUMANODES][STARPU_MAXCUDADEVS+STARPU_MAXNUMANODES]