T Linalg<T, H>::syr2k(
const T &a, const T &b, const value_type &alpha, const value_type &beta,
Uplo uplo) {
T c(a.allocator());
syr2k(a, b, &c, alpha, beta, uplo);
return c;
}
void MTLmarks::DmatDmatRun(std::string benchmark) {
if(benchmark == "dmatdmatadd"){
mtl_result = dmatdmatadd(size, steps);
}
else if(benchmark == "dmatdmatmult"){
mtl_result = dmatdmatmult(size, steps);
}
else if(benchmark == "cmajordmdmmult"){
mtl_result = cmajordmdmmult(size, steps);
}
else if(benchmark == "rmajordmdmmult"){
mtl_result = rmajordmdmmult(size, steps);
}
else if(benchmark == "nestedprod"){
mtl_result = nestedprod(size, steps);
}
else if(benchmark == "symm1"){
mtl_result = symm1(size, steps);
}
else if(benchmark == "symm1rect"){
mtl_result = symm1rect(size, steps);
}
else if(benchmark == "symm2"){
mtl_result = symm2(size, steps);
}
else if(benchmark == "syr2k"){
mtl_result = syr2k(size, steps);
}
else if(benchmark == "syr2krect"){
mtl_result = syr2krect(size, steps);
}
else if(benchmark == "syrk"){
mtl_result = syrk(size, steps);
}
else if(benchmark == "syrkrect"){
mtl_result = syrkrect(size, steps);
}
else if(benchmark == "custom"){
mtl_result = custom(size, steps);
}
else{
std::cerr << "MTLmarks benchmark does not exist." << std::endl;
exit(1);
}
}
int main(void)
{
double t_start, t_end;
DATA_TYPE* A;
DATA_TYPE* B;
DATA_TYPE* C;
DATA_TYPE* C_outputFromGpu;
A = (DATA_TYPE*)malloc(N*M*sizeof(DATA_TYPE));
B = (DATA_TYPE*)malloc(N*M*sizeof(DATA_TYPE));
C = (DATA_TYPE*)malloc(N*M*sizeof(DATA_TYPE));
C_outputFromGpu = (DATA_TYPE*)malloc(N*M*sizeof(DATA_TYPE));
init_arrays(A, B, C);
read_cl_file();
cl_initialization();
cl_mem_init(A, B, C);
cl_load_prog();
cl_launch_kernel();
errcode = clEnqueueReadBuffer(clCommandQue, c_mem_obj, CL_TRUE, 0, N*M*sizeof(DATA_TYPE), C_outputFromGpu, 0, NULL, NULL);
if(errcode != CL_SUCCESS) printf("Error in reading GPU mem\n");
t_start = rtclock();
syr2k(A, B, C);
t_end = rtclock();
fprintf(stdout, "CPU Runtime: %0.6lfs\n", t_end - t_start);
compareResults(C, C_outputFromGpu);
cl_clean_up();
free(A);
free(B);
free(C);
free(C_outputFromGpu);
return 0;
}
T& Linalg<T, H>::syr2k(
const T &a, const T &b, T &c, const value_type &alpha,
const value_type &beta, Uplo uplo) {
return const_cast< T& >(syr2k(
a, b, const_cast< const T& >(c), alpha, beta, uplo));
}
int main(void)
{
/* Prepare ctuning vars */
long ct_repeat=0;
long ct_repeat_max=1;
double t_start, t_end;
DATA_TYPE* A;
DATA_TYPE* B;
DATA_TYPE* C;
DATA_TYPE* C_outputFromGpu;
#ifdef OPENME
openme_init(NULL,NULL,NULL,0);
openme_callback("PROGRAM_START", NULL);
#endif
/* Run kernel. */
if (getenv("CT_REPEAT_MAIN")!=NULL) ct_repeat_max=atol(getenv("CT_REPEAT_MAIN"));
A = (DATA_TYPE*)malloc(N*M*sizeof(DATA_TYPE));
B = (DATA_TYPE*)malloc(N*M*sizeof(DATA_TYPE));
C = (DATA_TYPE*)malloc(N*M*sizeof(DATA_TYPE));
C_outputFromGpu = (DATA_TYPE*)malloc(N*M*sizeof(DATA_TYPE));
srand(1);
init_arrays(A, B, C);
read_cl_file();
cl_initialization();
cl_mem_init(A, B, C);
cl_load_prog();
#ifdef OPENME
openme_callback("ACC_KERNEL_START", NULL);
#endif
for (ct_repeat=0; ct_repeat<ct_repeat_max; ct_repeat++)
{
cl_launch_kernel();
err_code = clEnqueueReadBuffer(clCommandQue, c_mem_obj, CL_TRUE, 0, N*M*sizeof(DATA_TYPE), C_outputFromGpu, 0, NULL, NULL);
if(err_code != CL_SUCCESS)
{
printf("Error in reading GPU mem\n");
exit(1);
}
}
#ifdef OPENME
openme_callback("ACC_KERNEL_END", NULL);
#endif
srand(1);
init_arrays(A, B, C);
#ifdef OPENME
openme_callback("KERNEL_START", NULL);
#endif
for (ct_repeat=0; ct_repeat<ct_repeat_max; ct_repeat++)
{
syr2k(A, B, C);
}
#ifdef OPENME
openme_callback("KERNEL_END", NULL);
#endif
compareResults(C, C_outputFromGpu);
cl_clean_up();
free(A);
free(B);
free(C);
free(C_outputFromGpu);
#ifdef OPENME
openme_callback("PROGRAM_END", NULL);
#endif
return 0;
}
