oclutils.h File Reference

#include <string>
#include "gromacs/gpu_utils/gmxopencl.h"
#include "gromacs/utility/gmxassert.h"

Include dependency graph for oclutils.h:

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

Description

Declare utility routines for OpenCL.

Author

Anca Hamuraru anca@.nosp@m.stre.nosp@m.amcom.nosp@m.puti.nosp@m.ng.eu

Classes
struct	ocl_gpu_id_t
	OpenCL GPU device identificator. More...
struct	gmx_device_info_t
	OpenCL device information. More...
struct	gmx_device_runtime_data_t
	OpenCL GPU runtime data. More...

Enumerations
enum	ocl_vendor_id_t { OCL_VENDOR_NVIDIA = 0, OCL_VENDOR_AMD, OCL_VENDOR_INTEL, OCL_VENDOR_UNKNOWN }
	OpenCL vendor IDs.

Functions
int	ocl_copy_D2H (void *h_dest, cl_mem d_src, size_t offset, size_t bytes, GpuApiCallBehavior transferKind, cl_command_queue command_queue, cl_event *copy_event)
	Launches synchronous or asynchronous device to host memory copy. More...
int	ocl_copy_D2H_async (void *h_dest, cl_mem d_src, size_t offset, size_t bytes, cl_command_queue command_queue, cl_event *copy_event)
	Launches asynchronous device to host memory copy. More...
int	ocl_copy_H2D (cl_mem d_dest, void *h_src, size_t offset, size_t bytes, GpuApiCallBehavior transferKind, cl_command_queue command_queue, cl_event *copy_event)
	Launches synchronous or asynchronous host to device memory copy. More...
int	ocl_copy_H2D_async (cl_mem d_dest, void *h_src, size_t offset, size_t bytes, cl_command_queue command_queue, cl_event *copy_event)
	Launches asynchronous host to device memory copy. More...
int	ocl_copy_H2D_sync (cl_mem d_dest, void *h_src, size_t offset, size_t bytes, cl_command_queue command_queue)
	Launches synchronous host to device memory copy.
void	ocl_pmalloc (void **h_ptr, size_t nbytes)
	Allocate host memory in malloc style. More...
void	ocl_pfree (void *h_ptr)
	Free host memory in malloc style. More...
std::string	ocl_get_error_string (cl_int error)
	Convert error code to diagnostic string.
static void	gpuStreamSynchronize (cl_command_queue s)
	Calls clFinish() in the stream s. More...
static bool	haveStreamTasksCompleted (cl_command_queue s)
	Pretend to synchronize an OpenCL stream (dummy implementation). More...

Function Documentation

static void gpuStreamSynchronize ( cl_command_queue  s )

inlinestatic

Calls clFinish() in the stream s.

Parameters

[in]

s

stream to synchronize with

static bool haveStreamTasksCompleted ( cl_command_queue  s )

inlinestatic

Pretend to synchronize an OpenCL stream (dummy implementation).

Parameters

[in]

s

queue to check

Returns

True if all tasks enqueued in the stream s (at the time of this call) have completed.

int ocl_copy_D2H	(	void *	h_dest,
		cl_mem	d_src,
		size_t	offset,
		size_t	bytes,
		GpuApiCallBehavior	transferKind,
		cl_command_queue	command_queue,
		cl_event *	copy_event
	)

Launches synchronous or asynchronous device to host memory copy.

If copy_event is not NULL, on return it will contain an event object identifying this particular device to host operation. The event can further be used to queue a wait for this operation or to query profiling information.

int ocl_copy_D2H_async	(	void *	h_dest,
		cl_mem	d_src,
		size_t	offset,
		size_t	bytes,
		cl_command_queue	command_queue,
		cl_event *	copy_event
	)

Launches asynchronous device to host memory copy.

If copy_event is not NULL, on return it will contain an event object identifying this particular host to device operation. The event can further be used to queue a wait for this operation or to query profiling information.

int ocl_copy_H2D	(	cl_mem	d_dest,
		void *	h_src,
		size_t	offset,
		size_t	bytes,
		GpuApiCallBehavior	transferKind,
		cl_command_queue	command_queue,
		cl_event *	copy_event
	)

Launches synchronous or asynchronous host to device memory copy.

If copy_event is not NULL, on return it will contain an event object identifying this particular host to device operation. The event can further be used to queue a wait for this operation or to query profiling information.

int ocl_copy_H2D_async	(	cl_mem	d_dest,
		void *	h_src,
		size_t	offset,
		size_t	bytes,
		cl_command_queue	command_queue,
		cl_event *	copy_event
	)

Launches asynchronous host to device memory copy.

If copy_event is not NULL, on return it will contain an event object identifying this particular host to device operation. The event can further be used to queue a wait for this operation or to query profiling information.

void ocl_pfree

(

void *

h_ptr

)

Free host memory in malloc style.

Free host memory in malloc style.

Parameters

[in]

h_ptr

Buffer allocated with ocl_pmalloc that needs to be freed.

void ocl_pmalloc	(	void **	h_ptr,
		size_t	nbytes
	)

Allocate host memory in malloc style.

Allocate host memory in malloc style.

Todo:

This function should allocate page-locked memory to help reduce D2H and H2D transfer times, similar with pmalloc from pmalloc_cuda.cu.

Parameters

[in,out]	h_ptr	Pointer where to store the address of the newly allocated buffer.
[in]	nbytes	Size in bytes of the buffer to be allocated.