Improve ClBlast implementation, avoid recreating buffers, remove redundant transfers
This commit is contained in:
parent
f4257a8eef
commit
c7e5c4f7b2
1 changed files with 63 additions and 30 deletions
|
@ -18,36 +18,45 @@ cl_platform_id platform;
|
||||||
cl_device_id device;
|
cl_device_id device;
|
||||||
cl_context context;
|
cl_context context;
|
||||||
cl_command_queue queue;
|
cl_command_queue queue;
|
||||||
cl_event event;
|
|
||||||
bool cl_initialized = false;
|
bool cl_initialized = false;
|
||||||
|
size_t cl_size_a = 0, cl_size_b = 0, cl_size_c = 0;
|
||||||
|
cl_mem cl_buffer_a, cl_buffer_b, cl_buffer_c;
|
||||||
|
|
||||||
static void ggml_cl_sgemm_wrapper(const enum CBLAS_ORDER order, const enum CBLAS_TRANSPOSE trans_a, const enum CBLAS_TRANSPOSE trans_b, const int m, const int n, const int k, const float alpha, const float *host_a, const int lda, const float *host_b, const int ldb, const float beta, float *host_c, const int ldc) {
|
static void ggml_cl_sgemm_wrapper(const enum CBLAS_ORDER order, const enum CBLAS_TRANSPOSE trans_a, const enum CBLAS_TRANSPOSE trans_b, const int m, const int n, const int k, const float alpha, const float *host_a, const int lda, const float *host_b, const int ldb, const float beta, float *host_c, const int ldc) {
|
||||||
cl_int err = 0;
|
cl_int err = 0;
|
||||||
|
|
||||||
|
cl_event events[2];
|
||||||
|
events[0] = NULL;
|
||||||
|
events[1] = NULL;
|
||||||
|
|
||||||
if (!cl_initialized) {
|
if (!cl_initialized) {
|
||||||
char * KCPP_CLBLAST_PLATFORM = getenv("KCPP_CLBLAST_PLATFORM");
|
char * KCPP_CLBLAST_PLATFORM = getenv("KCPP_CLBLAST_PLATFORM");
|
||||||
char * KCPP_CLBLAST_DEVICES = getenv("KCPP_CLBLAST_DEVICES");
|
char * KCPP_CLBLAST_DEVICES = getenv("KCPP_CLBLAST_DEVICES");
|
||||||
int plat_num = (KCPP_CLBLAST_PLATFORM == NULL ? 0 : atoi(KCPP_CLBLAST_PLATFORM));
|
int plat_num = (KCPP_CLBLAST_PLATFORM == NULL ? 0 : atoi(KCPP_CLBLAST_PLATFORM));
|
||||||
int dev_num = (KCPP_CLBLAST_DEVICES == NULL ? 0 : atoi(KCPP_CLBLAST_DEVICES));
|
int dev_num = (KCPP_CLBLAST_DEVICES == NULL ? 0 : atoi(KCPP_CLBLAST_DEVICES));
|
||||||
printf("\nInitializing CLBlast (First Run)...");
|
printf("\nInitializing CLBlast (First Run)...");
|
||||||
printf("\nSelected: Platform=%d, Device=%d (If invalid, program will crash)\n",plat_num,dev_num);
|
printf("\nAttempting to use: Platform=%d, Device=%d (If invalid, program will crash)\n",plat_num,dev_num);
|
||||||
cl_uint num_platforms;
|
cl_uint num_platforms;
|
||||||
clGetPlatformIDs(0, NULL, &num_platforms);
|
clGetPlatformIDs(0, NULL, &num_platforms);
|
||||||
cl_platform_id* platforms = (cl_platform_id*)malloc(num_platforms*sizeof(cl_platform_id));
|
cl_platform_id* platforms = (cl_platform_id*)malloc(num_platforms*sizeof(cl_platform_id));
|
||||||
clGetPlatformIDs(num_platforms, platforms, NULL);
|
clGetPlatformIDs(num_platforms, platforms, NULL);
|
||||||
platform = platforms[plat_num];
|
platform = platforms[plat_num];
|
||||||
|
char platform_buffer[1024];
|
||||||
|
clGetPlatformInfo(platform, CL_PLATFORM_NAME, sizeof(platform_buffer), &platform_buffer, NULL);
|
||||||
cl_uint num_devices;
|
cl_uint num_devices;
|
||||||
clGetDeviceIDs(platform, CL_DEVICE_TYPE_ALL, 0, NULL, &num_devices);
|
clGetDeviceIDs(platform, CL_DEVICE_TYPE_ALL, 0, NULL, &num_devices);
|
||||||
cl_device_id* devices = (cl_device_id*)malloc(num_devices*sizeof(cl_device_id));
|
cl_device_id* devices = (cl_device_id*)malloc(num_devices*sizeof(cl_device_id));
|
||||||
clGetDeviceIDs(platform, CL_DEVICE_TYPE_ALL, num_devices, devices, NULL);
|
clGetDeviceIDs(platform, CL_DEVICE_TYPE_ALL, num_devices, devices, NULL);
|
||||||
device = devices[dev_num];
|
device = devices[dev_num];
|
||||||
|
char device_buffer[1024];
|
||||||
|
clGetDeviceInfo(device, CL_DEVICE_NAME, sizeof(device_buffer), &device_buffer, NULL);
|
||||||
|
printf("Using Platform: %s Device: %s\n", platform_buffer, device_buffer);
|
||||||
context = clCreateContext(NULL, 1, &device, NULL, NULL, &err);
|
context = clCreateContext(NULL, 1, &device, NULL, NULL, &err);
|
||||||
if (err != CL_SUCCESS) {
|
if (err != CL_SUCCESS) {
|
||||||
printf("Error creating OpenCL context: %d\n", err);
|
printf("Error creating OpenCL context: %d\n", err);
|
||||||
fflush(stdout);
|
fflush(stdout);
|
||||||
}
|
}
|
||||||
queue = clCreateCommandQueue(context, device, 0, &err);
|
queue = clCreateCommandQueue(context, device, 0, &err);
|
||||||
event = NULL;
|
|
||||||
|
|
||||||
if (err != CL_SUCCESS) {
|
if (err != CL_SUCCESS) {
|
||||||
printf("Error creating OpenCL Command Queue: %d\n", err);
|
printf("Error creating OpenCL Command Queue: %d\n", err);
|
||||||
|
@ -56,29 +65,56 @@ static void ggml_cl_sgemm_wrapper(const enum CBLAS_ORDER order, const enum CBLAS
|
||||||
|
|
||||||
free(platforms);
|
free(platforms);
|
||||||
free(devices);
|
free(devices);
|
||||||
|
|
||||||
|
cl_size_a = m * k * sizeof(float);
|
||||||
|
cl_size_b = n * k * sizeof(float);
|
||||||
|
cl_size_c = m * n * sizeof(float);
|
||||||
|
|
||||||
|
// Prepare buffers
|
||||||
|
cl_buffer_a = clCreateBuffer(context, CL_MEM_READ_ONLY, cl_size_a, NULL, &err);
|
||||||
|
if (err != CL_SUCCESS) {
|
||||||
|
printf("Error creating OpenCL Buffer A: %d\n", err);
|
||||||
|
fflush(stdout);
|
||||||
|
}
|
||||||
|
cl_buffer_b = clCreateBuffer(context, CL_MEM_READ_ONLY, cl_size_b, NULL, &err);
|
||||||
|
if (err != CL_SUCCESS) {
|
||||||
|
printf("Error creating OpenCL Buffer B: %d\n", err);
|
||||||
|
fflush(stdout);
|
||||||
|
}
|
||||||
|
cl_buffer_c = clCreateBuffer(context, CL_MEM_READ_WRITE, cl_size_c, NULL, &err);
|
||||||
|
if (err != CL_SUCCESS) {
|
||||||
|
printf("Error creating OpenCL Buffer C: %d\n", err);
|
||||||
|
fflush(stdout);
|
||||||
|
}
|
||||||
|
|
||||||
cl_initialized = true;
|
cl_initialized = true;
|
||||||
}
|
}
|
||||||
|
|
||||||
// Prepare buffers
|
// Resize buffers if too small
|
||||||
cl_mem cl_buffer_a = clCreateBuffer(context, CL_MEM_READ_WRITE, m*k*sizeof(float), NULL, &err);
|
if (m * k * sizeof(float) > cl_size_a) {
|
||||||
if (err != CL_SUCCESS) {
|
clReleaseMemObject(cl_buffer_a);
|
||||||
printf("Error creating OpenCL Buffer A: %d\n", err);
|
cl_size_a = m * k * sizeof(float);
|
||||||
fflush(stdout);
|
cl_buffer_a = clCreateBuffer(context, CL_MEM_READ_ONLY, cl_size_a, NULL, NULL);
|
||||||
}
|
}
|
||||||
cl_mem cl_buffer_b = clCreateBuffer(context, CL_MEM_READ_WRITE, n*k*sizeof(float), NULL, &err);
|
if (n * k * sizeof(float) > cl_size_b) {
|
||||||
if (err != CL_SUCCESS) {
|
clReleaseMemObject(cl_buffer_b);
|
||||||
printf("Error creating OpenCL Buffer B: %d\n", err);
|
cl_size_b = n * k * sizeof(float);
|
||||||
fflush(stdout);
|
cl_buffer_b = clCreateBuffer(context, CL_MEM_READ_ONLY, cl_size_b, NULL, NULL);
|
||||||
}
|
}
|
||||||
cl_mem cl_buffer_c = clCreateBuffer(context, CL_MEM_READ_WRITE, m*n*sizeof(float), NULL, &err);
|
if (m * n * sizeof(float) > cl_size_c) {
|
||||||
if (err != CL_SUCCESS) {
|
clReleaseMemObject(cl_buffer_c);
|
||||||
printf("Error creating OpenCL Buffer C: %d\n", err);
|
cl_size_c = m * n * sizeof(float);
|
||||||
fflush(stdout);
|
cl_buffer_c = clCreateBuffer(context, CL_MEM_READ_WRITE, cl_size_c, NULL, NULL);
|
||||||
}
|
}
|
||||||
|
|
||||||
clEnqueueWriteBuffer(queue, cl_buffer_a, CL_TRUE, 0, m*k*sizeof(float), host_a, 0, NULL, NULL);
|
clEnqueueWriteBuffer(queue, cl_buffer_a, CL_TRUE, 0, cl_size_a, host_a, 0, NULL, events);
|
||||||
clEnqueueWriteBuffer(queue, cl_buffer_b, CL_TRUE, 0, n*k*sizeof(float), host_b, 0, NULL, NULL);
|
clEnqueueWriteBuffer(queue, cl_buffer_b, CL_TRUE, 0, cl_size_b, host_b, 0, NULL, events + 1);
|
||||||
clEnqueueWriteBuffer(queue, cl_buffer_c, CL_TRUE, 0, m*n*sizeof(float), host_c, 0, NULL, NULL);
|
// buffer c is not required for this use case
|
||||||
|
// clEnqueueWriteBuffer(queue, cl_buffer_c, CL_TRUE, 0, cl_size_c, host_c, 0, NULL, NULL);
|
||||||
|
|
||||||
|
clWaitForEvents(2, events);
|
||||||
|
clReleaseEvent(events[0]);
|
||||||
|
clReleaseEvent(events[1]);
|
||||||
|
|
||||||
// Call the SGEMM routine.
|
// Call the SGEMM routine.
|
||||||
CLBlastStatusCode status = CLBlastSgemm(order,
|
CLBlastStatusCode status = CLBlastSgemm(order,
|
||||||
|
@ -89,19 +125,16 @@ static void ggml_cl_sgemm_wrapper(const enum CBLAS_ORDER order, const enum CBLAS
|
||||||
cl_buffer_b, 0, ldb,
|
cl_buffer_b, 0, ldb,
|
||||||
beta,
|
beta,
|
||||||
cl_buffer_c, 0, ldc,
|
cl_buffer_c, 0, ldc,
|
||||||
&queue, &event);
|
&queue, events);
|
||||||
|
|
||||||
|
clEnqueueReadBuffer(queue, cl_buffer_c, CL_TRUE, 0, m*n*sizeof(float), host_c, 1, events, events + 1);
|
||||||
|
|
||||||
// Wait for completion
|
// Wait for completion
|
||||||
if (status == CLBlastSuccess) {
|
if (status == CLBlastSuccess) {
|
||||||
clWaitForEvents(1, &event);
|
clWaitForEvents(2, events);
|
||||||
clReleaseEvent(event);
|
clReleaseEvent(events[0]);
|
||||||
|
clReleaseEvent(events[1]);
|
||||||
}
|
}
|
||||||
|
|
||||||
clEnqueueReadBuffer(queue, cl_buffer_c, CL_TRUE, 0, m*n*sizeof(float), host_c, 0, NULL, NULL);
|
|
||||||
|
|
||||||
clReleaseMemObject(cl_buffer_a);
|
|
||||||
clReleaseMemObject(cl_buffer_b);
|
|
||||||
clReleaseMemObject(cl_buffer_c);
|
|
||||||
}
|
}
|
||||||
|
|
||||||
#endif
|
#endif
|
||||||
|
@ -117,4 +150,4 @@ ggml_cl_sgemm_wrapper(Order, TransA, TransB, M, N, K, alpha, A, lda, B, ldb, bet
|
||||||
cblas_sgemm(Order, TransA, TransB, M, N, K, alpha, A, lda, B, ldb, beta, C, ldc);\
|
cblas_sgemm(Order, TransA, TransB, M, N, K, alpha, A, lda, B, ldb, beta, C, ldc);\
|
||||||
})
|
})
|
||||||
#endif
|
#endif
|
||||||
#endif
|
#endif
|
||||||
|
|
Loading…
Add table
Add a link
Reference in a new issue