int main(int argc, char** argv)
{
/* Retrieve problem size. */
int n = N;
int tsteps = TSTEPS;
/* Variable declaration/allocation. */
POLYBENCH_1D_ARRAY_DECL(A, DATA_TYPE, N, n);
POLYBENCH_1D_ARRAY_DECL(B, DATA_TYPE, N, n);
/* Initialize array(s). */
init_array (n, POLYBENCH_ARRAY(A), POLYBENCH_ARRAY(B));
/* Start timer. */
polybench_start_instruments;
/* Run kernel. */
kernel_jacobi_1d_imper (tsteps, n, POLYBENCH_ARRAY(A), POLYBENCH_ARRAY(B));
/* Stop and print timer. */
polybench_stop_instruments;
polybench_print_instruments;
/* Prevent dead-code elimination. All live-out data must be printed
by the function call in argument. */
polybench_prevent_dce(print_array(n, POLYBENCH_ARRAY(A)));
/* Be clean. */
POLYBENCH_FREE_ARRAY(A);
POLYBENCH_FREE_ARRAY(B);
return 0;
}
int main(int argc, char** argv)
{
/* Retrieve problem size. */
int n = N;
/* Variable declaration/allocation. */
POLYBENCH_2D_ARRAY_DECL(L, DATA_TYPE, N, N, n, n);
POLYBENCH_1D_ARRAY_DECL(x, DATA_TYPE, N, n);
POLYBENCH_1D_ARRAY_DECL(b, DATA_TYPE, N, n);
/* Initialize array(s). */
init_array (n, POLYBENCH_ARRAY(L), POLYBENCH_ARRAY(x), POLYBENCH_ARRAY(b));
/* Start timer. */
polybench_start_instruments;
/* Run kernel. */
kernel_trisolv (n, POLYBENCH_ARRAY(L), POLYBENCH_ARRAY(x), POLYBENCH_ARRAY(b));
/* Stop and print timer. */
polybench_stop_instruments;
polybench_print_instruments;
/* Prevent dead-code elimination. All live-out data must be printed
by the function call in argument. */
polybench_prevent_dce(print_array(n, POLYBENCH_ARRAY(x)));
/* Be clean. */
POLYBENCH_FREE_ARRAY(L);
POLYBENCH_FREE_ARRAY(x);
POLYBENCH_FREE_ARRAY(b);
return 0;
}
int main(int argc, char** argv)
{
#pragma hmpp atax acquire
/* Retrieve problem size. */
int nx = NX;
int ny = NY;
/* Variable declaration/allocation. */
POLYBENCH_2D_ARRAY_DECL(A, DATA_TYPE, NX, NY, nx, ny);
POLYBENCH_1D_ARRAY_DECL(x, DATA_TYPE, NY, ny);
POLYBENCH_1D_ARRAY_DECL(y, DATA_TYPE, NY, ny);
POLYBENCH_1D_ARRAY_DECL(tmp, DATA_TYPE, NX, nx);
#pragma hmpp atax allocate, &
#pragma hmpp & args[A].size={nx,ny}, args[A].hostdata="A", &
#pragma hmpp & args[x].size={ny}, args[x].hostdata="x", &
#pragma hmpp & args[y].size={ny}, args[y].hostdata="y", &
#pragma hmpp & args[tmp].size={nx}, args[tmp].hostdata="tmp"
/* Initialize array(s). */
init_array (nx, ny, POLYBENCH_ARRAY(A), POLYBENCH_ARRAY(x));
#pragma hmpp atax advancedload, args[A;x]
/* Start timer. */
polybench_start_instruments;
/* Run kernel. */
#pragma hmpp atax callsite
kernel_atax (nx, ny,
POLYBENCH_ARRAY(A),
POLYBENCH_ARRAY(x),
POLYBENCH_ARRAY(y),
POLYBENCH_ARRAY(tmp));
/* Stop and print timer. */
polybench_stop_instruments;
polybench_print_instruments;
#pragma hmpp atax delegatedstore, args[y]
/* Prevent dead-code elimination. All live-out data must be printed
by the function call in argument. */
polybench_prevent_dce(print_array(nx, POLYBENCH_ARRAY(y)));
/* Be clean. */
POLYBENCH_FREE_ARRAY(A);
POLYBENCH_FREE_ARRAY(x);
POLYBENCH_FREE_ARRAY(y);
POLYBENCH_FREE_ARRAY(tmp);
#pragma hmpp atax release
return 0;
}
int main(int argc, char** argv)
{
/* Retrieve problem size. */
int q = Q;
int d = D;
/* Variable declaration/allocation. */
POLYBENCH_2D_ARRAY_DECL(M, DATA_TYPE, Q, D, q, d);
POLYBENCH_2D_ARRAY_DECL(MatchQ, DATA_TYPE, Q, D, q, d);
POLYBENCH_1D_ARRAY_DECL(QS, DATA_TYPE, Q, q);
POLYBENCH_1D_ARRAY_DECL(DB, DATA_TYPE, D, d);
/* Initialize array(s). */
init_array (q, d,
POLYBENCH_ARRAY(QS),
POLYBENCH_ARRAY(DB),
POLYBENCH_ARRAY(MatchQ));
/* Start timer. */
polybench_start_instruments;
/*
DATA_TYPE POLYBENCH_1D(QS,Q,q),
DATA_TYPE POLYBENCH_1D(DB,D,d),
DATA_TYPE POLYBENCH_2D(MatchQ,Q,D,q,d),
DATA_TYPE POLYBENCH_2D(M,Q,D,q,d))
*/
/* Run kernel. */
kernel_mvt (q, d,
POLYBENCH_ARRAY(QS),
POLYBENCH_ARRAY(DB),
POLYBENCH_ARRAY(MatchQ),
POLYBENCH_ARRAY(M));
/* Stop and print timer. */
polybench_stop_instruments;
polybench_print_instruments;
/* Prevent dead-code elimination. All live-out data must be printed
by the function call in argument. */
polybench_prevent_dce(print_array(q, d,
POLYBENCH_ARRAY(QS),
POLYBENCH_ARRAY(DB),
POLYBENCH_ARRAY(MatchQ)));
/* Be clean. */
POLYBENCH_FREE_ARRAY(M);
POLYBENCH_FREE_ARRAY(QS);
POLYBENCH_FREE_ARRAY(DB);
POLYBENCH_FREE_ARRAY(MatchQ);
return 0;
}
int main(void)
{
int nx = NX;
int ny = NY;
POLYBENCH_2D_ARRAY_DECL(A,DATA_TYPE,NX,NY,nx,ny);
POLYBENCH_1D_ARRAY_DECL(x,DATA_TYPE,NY,ny);
POLYBENCH_1D_ARRAY_DECL(y,DATA_TYPE,NY,ny);
POLYBENCH_1D_ARRAY_DECL(y_outputFromGpu,DATA_TYPE,NY,ny);
POLYBENCH_1D_ARRAY_DECL(tmp,DATA_TYPE,NX,nx);
init_array(nx, ny, POLYBENCH_ARRAY(x), POLYBENCH_ARRAY(A));
read_cl_file();
cl_initialization();
cl_mem_init(POLYBENCH_ARRAY(A), POLYBENCH_ARRAY(x), POLYBENCH_ARRAY(y), POLYBENCH_ARRAY(tmp));
cl_load_prog();
cl_launch_kernel(nx, ny);
errcode = clEnqueueReadBuffer(clCommandQue, y_mem_obj, CL_TRUE, 0, NY*sizeof(DATA_TYPE), POLYBENCH_ARRAY(y_outputFromGpu), 0, NULL, NULL);
if(errcode != CL_SUCCESS) printf("Error in reading GPU mem\n");
#ifdef RUN_ON_CPU
/* Start timer. */
polybench_start_instruments;
atax_cpu(nx, ny, POLYBENCH_ARRAY(A), POLYBENCH_ARRAY(x), POLYBENCH_ARRAY(y), POLYBENCH_ARRAY(tmp));
/* Stop and print timer. */
printf("CPU Time in seconds:\n");
polybench_stop_instruments;
polybench_print_instruments;
compareResults(ny, POLYBENCH_ARRAY(y), POLYBENCH_ARRAY(y_outputFromGpu));
#else
print_array(ny, POLYBENCH_ARRAY(y_outputFromGpu));
#endif //RUN_ON_CPU
cl_clean_up();
POLYBENCH_FREE_ARRAY(A);
POLYBENCH_FREE_ARRAY(x);
POLYBENCH_FREE_ARRAY(y);
POLYBENCH_FREE_ARRAY(y_outputFromGpu);
POLYBENCH_FREE_ARRAY(tmp);
return 0;
}
int main(int argc, char** argv)
{
/* Retrieve problem size. */
int n = N;
/* Variable declaration/allocation. */
POLYBENCH_2D_ARRAY_DECL(y, DATA_TYPE, N, N, n, n);
POLYBENCH_2D_ARRAY_DECL(sum, DATA_TYPE, N, N, n, n);
POLYBENCH_1D_ARRAY_DECL(alpha, DATA_TYPE, N, n);
POLYBENCH_1D_ARRAY_DECL(beta, DATA_TYPE, N, n);
POLYBENCH_1D_ARRAY_DECL(r, DATA_TYPE, N, n);
POLYBENCH_1D_ARRAY_DECL(out, DATA_TYPE, N, n);
/* Initialize array(s). */
init_array (n,
POLYBENCH_ARRAY(y),
POLYBENCH_ARRAY(sum),
POLYBENCH_ARRAY(alpha),
POLYBENCH_ARRAY(beta),
POLYBENCH_ARRAY(r));
/* Start timer. */
polybench_start_instruments;
/* Run kernel. */
kernel_durbin (n,
POLYBENCH_ARRAY(y),
POLYBENCH_ARRAY(sum),
POLYBENCH_ARRAY(alpha),
POLYBENCH_ARRAY(beta),
POLYBENCH_ARRAY(r),
POLYBENCH_ARRAY(out));
/* Stop and print timer. */
polybench_stop_instruments;
polybench_print_instruments;
/* Prevent dead-code elimination. All live-out data must be printed
by the function call in argument. */
polybench_prevent_dce(print_array(n, POLYBENCH_ARRAY(out)));
/* Be clean. */
POLYBENCH_FREE_ARRAY(y);
POLYBENCH_FREE_ARRAY(sum);
POLYBENCH_FREE_ARRAY(alpha);
POLYBENCH_FREE_ARRAY(beta);
POLYBENCH_FREE_ARRAY(r);
POLYBENCH_FREE_ARRAY(out);
return 0;
}
int main(int argc, char** argv)
{
/* Retrieve problem size. */
int n = N;
/* Variable declaration/allocation. */
DATA_TYPE alpha;
DATA_TYPE beta;
POLYBENCH_2D_ARRAY_DECL(A, DATA_TYPE, N, N, n, n);
POLYBENCH_2D_ARRAY_DECL(B, DATA_TYPE, N, N, n, n);
POLYBENCH_1D_ARRAY_DECL(tmp, DATA_TYPE, N, n);
POLYBENCH_1D_ARRAY_DECL(x, DATA_TYPE, N, n);
POLYBENCH_1D_ARRAY_DECL(y, DATA_TYPE, N, n);
/* Initialize array(s). */
init_array (n, &alpha, &beta,
POLYBENCH_ARRAY(A),
POLYBENCH_ARRAY(B),
POLYBENCH_ARRAY(x));
/* Start timer. */
polybench_start_instruments;
/* Run kernel. */
kernel_gesummv (n, alpha, beta,
POLYBENCH_ARRAY(A),
POLYBENCH_ARRAY(B),
POLYBENCH_ARRAY(tmp),
POLYBENCH_ARRAY(x),
POLYBENCH_ARRAY(y));
/* Stop and print timer. */
polybench_stop_instruments;
polybench_print_instruments;
/* Prevent dead-code elimination. All live-out data must be printed
by the function call in argument. */
polybench_prevent_dce(print_array(n, POLYBENCH_ARRAY(y)));
/* Be clean. */
POLYBENCH_FREE_ARRAY(A);
POLYBENCH_FREE_ARRAY(B);
POLYBENCH_FREE_ARRAY(tmp);
POLYBENCH_FREE_ARRAY(x);
POLYBENCH_FREE_ARRAY(y);
return 0;
}
int main(int argc, char** argv)
{
if (argc < 2) {
printf("Parametros: [nThreads]\n");
exit(1);
}
omp_set_num_threads(atoi(argv[1]));
/* Retrieve problem size. */
int n = N;
int m = M;
/* Variable declaration/allocation. */
DATA_TYPE float_n;
POLYBENCH_2D_ARRAY_DECL(data,DATA_TYPE,M,N,m,n);
POLYBENCH_2D_ARRAY_DECL(symmat,DATA_TYPE,M,M,m,m);
POLYBENCH_1D_ARRAY_DECL(mean,DATA_TYPE,M,m);
POLYBENCH_1D_ARRAY_DECL(stddev,DATA_TYPE,M,m);
/* Initialize array(s). */
init_array (m, n, &float_n, POLYBENCH_ARRAY(data));
/* Start timer. */
polybench_start_instruments;
/* Run kernel. */
kernel_correlation (m, n, float_n,
POLYBENCH_ARRAY(data),
POLYBENCH_ARRAY(symmat),
POLYBENCH_ARRAY(mean),
POLYBENCH_ARRAY(stddev));
/* Stop and print timer. */
polybench_stop_instruments;
polybench_print_instruments;
/* Prevent dead-code elimination. All live-out data must be printed
by the function call in argument. */
polybench_prevent_dce(print_array(m, POLYBENCH_ARRAY(symmat)));
/* Be clean. */
POLYBENCH_FREE_ARRAY(data);
POLYBENCH_FREE_ARRAY(symmat);
POLYBENCH_FREE_ARRAY(mean);
POLYBENCH_FREE_ARRAY(stddev);
return 0;
}
int main(int argc, char** argv)
{
/* Retrieve problem size. */
int nx = NX;
int ny = NY;
/* Variable declaration/allocation. */
POLYBENCH_2D_ARRAY_DECL(A, DATA_TYPE, NX, NY, nx, ny);
POLYBENCH_1D_ARRAY_DECL(s, DATA_TYPE, NY, ny);
POLYBENCH_1D_ARRAY_DECL(q, DATA_TYPE, NX, nx);
POLYBENCH_1D_ARRAY_DECL(p, DATA_TYPE, NY, ny);
POLYBENCH_1D_ARRAY_DECL(r, DATA_TYPE, NX, nx);
/* Initialize array(s). */
init_array (nx, ny,
POLYBENCH_ARRAY(A),
POLYBENCH_ARRAY(r),
POLYBENCH_ARRAY(p));
/* Start timer. */
polybench_start_instruments;
/* Run kernel. */
kernel_bicg (nx, ny,
POLYBENCH_ARRAY(A),
POLYBENCH_ARRAY(s),
POLYBENCH_ARRAY(q),
POLYBENCH_ARRAY(p),
POLYBENCH_ARRAY(r));
/* Stop and print timer. */
polybench_stop_instruments;
polybench_print_instruments;
/* Prevent dead-code elimination. All live-out data must be printed
by the function call in argument. */
polybench_prevent_dce(print_array(nx, ny, POLYBENCH_ARRAY(s), POLYBENCH_ARRAY(q)));
/* Be clean. */
POLYBENCH_FREE_ARRAY(A);
POLYBENCH_FREE_ARRAY(s);
POLYBENCH_FREE_ARRAY(q);
POLYBENCH_FREE_ARRAY(p);
POLYBENCH_FREE_ARRAY(r);
return 0;
}
int main(int argc, char** argv)
{
/* Prepare ctuning vars */
long ct_repeat=0;
long ct_repeat_max=1;
int ct_return=0;
/* Retrieve problem size. */
int n = N;
#ifdef OPENME
openme_init(NULL,NULL,NULL,0);
openme_callback("PROGRAM_START", NULL);
#endif
/* Variable declaration/allocation. */
POLYBENCH_2D_ARRAY_DECL(A, DATA_TYPE, N, N, n, n);
POLYBENCH_1D_ARRAY_DECL(p, DATA_TYPE, N, n);
/* Initialize array(s). */
init_array (n, POLYBENCH_ARRAY(p), POLYBENCH_ARRAY(A));
/* Start timer. */
polybench_start_instruments;
/* Run kernel. */
if (getenv("CT_REPEAT_MAIN")!=NULL) ct_repeat_max=atol(getenv("CT_REPEAT_MAIN"));
#ifdef OPENME
openme_callback("KERNEL_START", NULL);
#endif
for (ct_repeat=0; ct_repeat<ct_repeat_max; ct_repeat++)
kernel_cholesky (n, POLYBENCH_ARRAY(p), POLYBENCH_ARRAY(A));
#ifdef OPENME
openme_callback("KERNEL_END", NULL);
#endif
/* Stop and print timer. */
polybench_stop_instruments;
polybench_print_instruments;
/* Prevent dead-code elimination. All live-out data must be printed
by the function call in argument. */
polybench_prevent_dce(print_array(n, POLYBENCH_ARRAY(A)));
/* Be clean. */
POLYBENCH_FREE_ARRAY(A);
POLYBENCH_FREE_ARRAY(p);
#ifdef OPENME
openme_callback("PROGRAM_END", NULL);
#endif
return 0;
}
int main(int argc, char *argv[])
{
int tmax = TMAX;
int nx = NX;
int ny = NY;
POLYBENCH_1D_ARRAY_DECL(_fict_,DATA_TYPE,TMAX,TMAX);
POLYBENCH_2D_ARRAY_DECL(ex,DATA_TYPE,NX,NY,nx,ny);
POLYBENCH_2D_ARRAY_DECL(ey,DATA_TYPE,NX,NY,nx,ny);
POLYBENCH_2D_ARRAY_DECL(hz,DATA_TYPE,NX,NY,nx,ny);
POLYBENCH_2D_ARRAY_DECL(hz_outputFromGpu,DATA_TYPE,NX,NY,nx,ny);
init_arrays(tmax, nx, ny, POLYBENCH_ARRAY(_fict_), POLYBENCH_ARRAY(ex), POLYBENCH_ARRAY(ey), POLYBENCH_ARRAY(hz));
read_cl_file();
cl_initialization();
cl_mem_init(POLYBENCH_ARRAY(_fict_), POLYBENCH_ARRAY(ex), POLYBENCH_ARRAY(ey), POLYBENCH_ARRAY(hz));
cl_load_prog();
cl_launch_kernel(tmax, nx, ny);
errcode = clEnqueueReadBuffer(clCommandQue, hz_mem_obj, CL_TRUE, 0, NX * NY * sizeof(DATA_TYPE), POLYBENCH_ARRAY(hz_outputFromGpu), 0, NULL, NULL);
if(errcode != CL_SUCCESS) printf("Error in reading GPU mem\n");
#ifdef RUN_ON_CPU
/* Start timer. */
polybench_start_instruments;
runFdtd(tmax, nx, ny, POLYBENCH_ARRAY(_fict_), POLYBENCH_ARRAY(ex), POLYBENCH_ARRAY(ey), POLYBENCH_ARRAY(hz));
/* Stop and print timer. */
printf("CPU Time in seconds:\n");
polybench_stop_instruments;
polybench_print_instruments;
compareResults(nx, ny, POLYBENCH_ARRAY(hz), POLYBENCH_ARRAY(hz_outputFromGpu));
#else //prevent dead code elimination
polybench_prevent_dce(print_array(nx, ny, POLYBENCH_ARRAY(hz_outputFromGpu)));
#endif //RUN_ON_CPU
POLYBENCH_FREE_ARRAY(_fict_);
POLYBENCH_FREE_ARRAY(ex);
POLYBENCH_FREE_ARRAY(ey);
POLYBENCH_FREE_ARRAY(hz);
POLYBENCH_FREE_ARRAY(hz_outputFromGpu);
cl_clean_up();
return 0;
}
int main(int argc, char** argv)
{
fid = fopen("memory.txt", "w+t");
/* Retrieve problem size. */
int tmax = TMAX;
int nx = NX;
int ny = NY;
/* Variable declaration/allocation. */
POLYBENCH_2D_ARRAY_DECL(ex,DATA_TYPE,NX,NY,nx,ny);
POLYBENCH_2D_ARRAY_DECL(ey,DATA_TYPE,NX,NY,nx,ny);
POLYBENCH_2D_ARRAY_DECL(hz,DATA_TYPE,NX,NY,nx,ny);
POLYBENCH_1D_ARRAY_DECL(_fict_,DATA_TYPE,TMAX,tmax);
/* Initialize array(s). */
init_array (tmax, nx, ny,
POLYBENCH_ARRAY(ex),
POLYBENCH_ARRAY(ey),
POLYBENCH_ARRAY(hz),
POLYBENCH_ARRAY(_fict_));
/* Start timer. */
polybench_start_instruments;
/* Run kernel. */
kernel_fdtd_2d (tmax, nx, ny,
POLYBENCH_ARRAY(ex),
POLYBENCH_ARRAY(ey),
POLYBENCH_ARRAY(hz),
POLYBENCH_ARRAY(_fict_));
/* Stop and print timer. */
polybench_stop_instruments;
polybench_print_instruments;
/* Prevent dead-code elimination. All live-out data must be printed
by the function call in argument. */
print_array(nx, ny, POLYBENCH_ARRAY(ex),
POLYBENCH_ARRAY(ey),
POLYBENCH_ARRAY(hz));
/* Be clean. */
POLYBENCH_FREE_ARRAY(ex);
POLYBENCH_FREE_ARRAY(ey);
POLYBENCH_FREE_ARRAY(hz);
POLYBENCH_FREE_ARRAY(_fict_);
return 0;
}
int main(int argc, char** argv)
{
fid = fopen("memory.txt", "w+t");
/* Retrieve problem size. */
int n = N;
int m = M;
/* Variable declaration/allocation. */
DATA_TYPE float_n;
POLYBENCH_2D_ARRAY_DECL(data,DATA_TYPE,M,N,m,n);
POLYBENCH_2D_ARRAY_DECL(symmat,DATA_TYPE,M,M,m,m);
POLYBENCH_1D_ARRAY_DECL(mean,DATA_TYPE,M,m);
/* Initialize array(s). */
init_array (m, n, &float_n, POLYBENCH_ARRAY(data));
/* Start timer. */
polybench_start_instruments;
/* Run kernel. */
kernel_covariance (m, n, float_n,
POLYBENCH_ARRAY(data),
POLYBENCH_ARRAY(symmat),
POLYBENCH_ARRAY(mean));
/* Stop and print timer. */
polybench_stop_instruments;
polybench_print_instruments;
/* Prevent dead-code elimination. All live-out data must be printed
by the function call in argument. */
polybench_prevent_dce(print_array(m, POLYBENCH_ARRAY(symmat)));
/* Be clean. */
POLYBENCH_FREE_ARRAY(data);
POLYBENCH_FREE_ARRAY(symmat);
POLYBENCH_FREE_ARRAY(mean);
return 0;
}
int main(int argc, char** argv)
{
/* Retrieve problem size. */
int nr = NR;
int nq = NQ;
int np = NP;
/* Variable declaration/allocation. */
POLYBENCH_3D_ARRAY_DECL(A,DATA_TYPE,NR,NQ,NP,nr,nq,np);
POLYBENCH_1D_ARRAY_DECL(sum,DATA_TYPE,NP,np);
POLYBENCH_2D_ARRAY_DECL(C4,DATA_TYPE,NP,NP,np,np);
/* Initialize array(s). */
init_array (nr, nq, np,
POLYBENCH_ARRAY(A),
POLYBENCH_ARRAY(C4));
/* Start timer. */
polybench_start_instruments;
/* Run kernel. */
kernel_doitgen (nr, nq, np,
POLYBENCH_ARRAY(A),
POLYBENCH_ARRAY(C4),
POLYBENCH_ARRAY(sum));
/* Stop and print timer. */
polybench_stop_instruments;
polybench_print_instruments;
/* Prevent dead-code elimination. All live-out data must be printed
by the function call in argument. */
polybench_prevent_dce(print_array(nr, nq, np, POLYBENCH_ARRAY(A)));
/* Be clean. */
POLYBENCH_FREE_ARRAY(A);
POLYBENCH_FREE_ARRAY(sum);
POLYBENCH_FREE_ARRAY(C4);
return 0;
}
int main(int argc, char** argv)
{
#pragma hmpp fdtd2d acquire
/* Retrieve problem size. */
int tmax = TMAX;
int nx = NX;
int ny = NY;
/* Variable declaration/allocation. */
POLYBENCH_2D_ARRAY_DECL(ex,DATA_TYPE,NX,NY,nx,ny);
POLYBENCH_2D_ARRAY_DECL(ey,DATA_TYPE,NX,NY,nx,ny);
POLYBENCH_2D_ARRAY_DECL(hz,DATA_TYPE,NX,NY,nx,ny);
POLYBENCH_1D_ARRAY_DECL(_fict_,DATA_TYPE,TMAX,tmax);
#pragma hmpp fdtd2d allocate, &
#pragma hmpp & args[ex].size={nx,ny}, args[ex].hostdata="ex", &
#pragma hmpp & args[ey].size={nx,ny}, args[ey].hostdata="ey", &
#pragma hmpp & args[hz].size={nx,ny}, args[hz].hostdata="hz", &
#pragma hmpp & args[_fict_].size={tmax}, args[_fict_].hostdata="_fict_"
/* Initialize array(s). */
init_array (tmax, nx, ny,
POLYBENCH_ARRAY(ex),
POLYBENCH_ARRAY(ey),
POLYBENCH_ARRAY(hz),
POLYBENCH_ARRAY(_fict_));
/* Start timer. */
polybench_start_instruments;
/* Run kernel. */
kernel_fdtd_2d (tmax, nx, ny,
POLYBENCH_ARRAY(ex),
POLYBENCH_ARRAY(ey),
POLYBENCH_ARRAY(hz),
POLYBENCH_ARRAY(_fict_));
#pragma hmpp fdtd2d advancedload, args[ex;ey;hz;_fict_]
/* Stop and print timer. */
polybench_stop_instruments;
polybench_print_instruments;
#pragma hmpp fdtd2d delegatedstore, args[ex;ey;hz]
/* Prevent dead-code elimination. All live-out data must be printed
by the function call in argument. */
polybench_prevent_dce(print_array(nx, ny, POLYBENCH_ARRAY(ex),
POLYBENCH_ARRAY(ey),
POLYBENCH_ARRAY(hz)));
/* Be clean. */
POLYBENCH_FREE_ARRAY(ex);
POLYBENCH_FREE_ARRAY(ey);
POLYBENCH_FREE_ARRAY(hz);
POLYBENCH_FREE_ARRAY(_fict_);
#pragma hmpp fdtd2d release
return 0;
}
int main(int argc, char** argv)
{
/* Retrieve problem size. */
int n = N;
int tsteps = TSTEPS;
POLYBENCH_1D_ARRAY_DECL(a,DATA_TYPE,N,n);
POLYBENCH_1D_ARRAY_DECL(b,DATA_TYPE,N,n);
POLYBENCH_1D_ARRAY_DECL(a_outputFromGpu,DATA_TYPE,N,n);
POLYBENCH_1D_ARRAY_DECL(b_outputFromGpu,DATA_TYPE,N,n);
init_array(n, POLYBENCH_ARRAY(a), POLYBENCH_ARRAY(b));
#if OPENCL_DEVICE_SELECTION!=CL_DEVICE_TYPE_ACCELERATOR
read_cl_file();
#else
if (argc != 2){
printf("%s <inputfile>\n", argv[0]);
return EXIT_FAILURE;
}
read_cl_file(argv);
#endif
cl_initialization();
cl_mem_init(POLYBENCH_ARRAY(a), POLYBENCH_ARRAY(b));
cl_load_prog();
/* Start timer. */
polybench_start_instruments;
int t;
//for (t = 0; t < _PB_TSTEPS ; t++)
for (t = 0; t < _PB_TSTEPS / TT; t++)
{
cl_launch_kernels(0);
}
/* Stop and print timer. */
#if OPENCL_DEVICE_SELECTION==CL_DEVICE_TYPE_CPU
printf("OpenCL-CPU Time in seconds: ");
#elif OPENCL_DEVICE_SELECTION==CL_DEVICE_TYPE_GPU
printf("OpenCL-GPU Time in seconds: ");
#else
printf("OpenCL-FPGA Time in seconds: ");
#endif
polybench_stop_instruments;
polybench_print_instruments;
errcode = clEnqueueReadBuffer(clCommandQue, a_mem_obj_k1, CL_TRUE, 0, N/6 * sizeof(DATA_TYPE), POLYBENCH_ARRAY(a_outputFromGpu), 0, NULL, NULL);
errcode = clEnqueueReadBuffer(clCommandQue, a_mem_obj_k2, CL_TRUE, 0, N/6 * sizeof(DATA_TYPE), POLYBENCH_ARRAY(a_outputFromGpu)+N*1/6, 0, NULL, NULL);
errcode = clEnqueueReadBuffer(clCommandQue, a_mem_obj_k3, CL_TRUE, 0, N/6 * sizeof(DATA_TYPE), POLYBENCH_ARRAY(a_outputFromGpu)+N*2/6, 0, NULL, NULL);
errcode = clEnqueueReadBuffer(clCommandQue, a_mem_obj_k4, CL_TRUE, 0, N/6 * sizeof(DATA_TYPE), POLYBENCH_ARRAY(a_outputFromGpu)+N*3/6, 0, NULL, NULL);
errcode = clEnqueueReadBuffer(clCommandQue, a_mem_obj_k5, CL_TRUE, 0, N/6 * sizeof(DATA_TYPE), POLYBENCH_ARRAY(a_outputFromGpu)+N*4/6, 0, NULL, NULL);
errcode = clEnqueueReadBuffer(clCommandQue, a_mem_obj_k6, CL_TRUE, 0, N/6 * sizeof(DATA_TYPE), POLYBENCH_ARRAY(a_outputFromGpu)+N*5/6, 0, NULL, NULL);
errcode = clEnqueueReadBuffer(clCommandQue, b_mem_obj_k1, CL_TRUE, 0, N/6 * sizeof(DATA_TYPE), POLYBENCH_ARRAY(b_outputFromGpu), 0, NULL, NULL);
errcode = clEnqueueReadBuffer(clCommandQue, b_mem_obj_k2, CL_TRUE, 0, N/6 * sizeof(DATA_TYPE), POLYBENCH_ARRAY(b_outputFromGpu)+N*1/6, 0, NULL, NULL);
errcode = clEnqueueReadBuffer(clCommandQue, b_mem_obj_k3, CL_TRUE, 0, N/6 * sizeof(DATA_TYPE), POLYBENCH_ARRAY(b_outputFromGpu)+N*2/6, 0, NULL, NULL);
errcode = clEnqueueReadBuffer(clCommandQue, b_mem_obj_k4, CL_TRUE, 0, N/6 * sizeof(DATA_TYPE), POLYBENCH_ARRAY(b_outputFromGpu)+N*3/6, 0, NULL, NULL);
errcode = clEnqueueReadBuffer(clCommandQue, b_mem_obj_k5, CL_TRUE, 0, N/6 * sizeof(DATA_TYPE), POLYBENCH_ARRAY(b_outputFromGpu)+N*4/6, 0, NULL, NULL);
errcode = clEnqueueReadBuffer(clCommandQue, b_mem_obj_k6, CL_TRUE, 0, N/6 * sizeof(DATA_TYPE), POLYBENCH_ARRAY(b_outputFromGpu)+N*5/6, 0, NULL, NULL);
if(errcode != CL_SUCCESS) printf("Error in reading GPU mem\n");
#ifdef RUN_ON_CPU
/* Start timer. */
polybench_start_instruments;
runJacobi1DCpu(tsteps, n, POLYBENCH_ARRAY(a), POLYBENCH_ARRAY(b));
/* Stop and print timer. */
printf("CPU Time in seconds: ");
polybench_stop_instruments;
polybench_print_instruments;
compareResults(n, POLYBENCH_ARRAY(a), POLYBENCH_ARRAY(a_outputFromGpu), POLYBENCH_ARRAY(b), POLYBENCH_ARRAY(b_outputFromGpu));
#else //prevent dead code elimination
polybench_prevent_dce(print_array(n, POLYBENCH_ARRAY(a_outputFromGpu)));
#endif //RUN_ON_CPU
cl_clean_up();
POLYBENCH_FREE_ARRAY(a);
POLYBENCH_FREE_ARRAY(b);
POLYBENCH_FREE_ARRAY(a_outputFromGpu);
POLYBENCH_FREE_ARRAY(b_outputFromGpu);
return 0;
}
int main(int argc, char** argv)
{
#pragma hmpp gesummv acquire
/* Retrieve problem size. */
int n = N;
/* Variable declaration/allocation. */
DATA_TYPE alpha;
DATA_TYPE beta;
POLYBENCH_2D_ARRAY_DECL(A, DATA_TYPE, N, N, n, n);
POLYBENCH_2D_ARRAY_DECL(B, DATA_TYPE, N, N, n, n);
POLYBENCH_1D_ARRAY_DECL(tmp, DATA_TYPE, N, n);
POLYBENCH_1D_ARRAY_DECL(x, DATA_TYPE, N, n);
POLYBENCH_1D_ARRAY_DECL(y, DATA_TYPE, N, n);
#pragma hmpp gesummv allocate, &
#pragma hmpp & args[A].size={n,n}, args[A].hostdata="A", &
#pragma hmpp & args[B].size={n,n}, args[B].hostdata="B", &
#pragma hmpp & args[x].size={n}, args[x].hostdata="x", &
#pragma hmpp & args[y].size={n}, args[y].hostdata="y", &
#pragma hmpp & args[tmp].size={n}, args[tmp].hostdata="tmp"
/* Initialize array(s). */
init_array (n, &alpha, &beta,
POLYBENCH_ARRAY(A),
POLYBENCH_ARRAY(B),
POLYBENCH_ARRAY(x));
#pragma hmpp gesummv advancedload, args[A;B;x]
/* Start timer. */
polybench_start_instruments;
/* Run kernel. */
#pragma hmpp gesummv callsite
kernel_gesummv (n, alpha, beta,
POLYBENCH_ARRAY(A),
POLYBENCH_ARRAY(B),
POLYBENCH_ARRAY(tmp),
POLYBENCH_ARRAY(x),
POLYBENCH_ARRAY(y));
/* Stop and print timer. */
polybench_stop_instruments;
polybench_print_instruments;
#pragma hmpp gesummv delegatedstore, args[y]
/* Prevent dead-code elimination. All live-out data must be printed
by the function call in argument. */
polybench_prevent_dce(print_array(n, POLYBENCH_ARRAY(y)));
/* Be clean. */
POLYBENCH_FREE_ARRAY(A);
POLYBENCH_FREE_ARRAY(B);
POLYBENCH_FREE_ARRAY(tmp);
POLYBENCH_FREE_ARRAY(x);
POLYBENCH_FREE_ARRAY(y);
#pragma hmpp gesummv release
return 0;
}
