97 lines
2.8 KiB
Plaintext
97 lines
2.8 KiB
Plaintext
|
/* Jacobi-3 program */
|
||
|
|
|
||
|
|
#include <math.h>
|
||
|
|
#include <stdio.h>
|
||
|
|
|
||
|
|
#define Max(a, b) ((a) > (b) ? (a) : (b))
|
||
|
|
|
||
|
|
#define L 384
|
||
|
|
#define ITMAX 100
|
||
|
|
|
||
|
|
int i, j, k, it;
|
||
|
|
float eps;
|
||
|
|
float MAXEPS = 0.5f;
|
||
|
|
|
||
|
|
/* 3D arrays block distributed along 3 dimensions */
|
||
|
|
#pragma dvm array distribute[block][block][block]
|
||
|
|
float A[L][L][L];
|
||
|
|
#pragma dvm array align([i][j][k] with A[i][j][k])
|
||
|
|
float B[L][L][L];
|
||
|
|
|
||
|
|
int main(int an, char **as)
|
||
|
|
{
|
||
|
|
double startt, endt;
|
||
|
|
#pragma dvm region
|
||
|
|
{
|
||
|
|
/* 3D parallel loop with base array A */
|
||
|
|
#pragma dvm parallel([i][j][k] on A[i][j][k]) cuda_block(32, 8)
|
||
|
|
for (i = 0; i < L; i++)
|
||
|
|
for (j = 0; j < L; j++)
|
||
|
|
for (k = 0; k < L; k++)
|
||
|
|
{
|
||
|
|
A[i][j][k] = 0;
|
||
|
|
if (i == 0 || j == 0 || k == 0 || i == L - 1 || j == L - 1 || k == L - 1)
|
||
|
|
B[i][j][k] = 0;
|
||
|
|
else
|
||
|
|
B[i][j][k] = 4 + i + j + k;
|
||
|
|
}
|
||
|
|
}
|
||
|
|
|
||
|
|
#ifdef _DVMH
|
||
|
|
dvmh_barrier();
|
||
|
|
startt = dvmh_wtime();
|
||
|
|
#else
|
||
|
|
startt = 0;
|
||
|
|
#endif
|
||
|
|
/* iteration loop */
|
||
|
|
for (it = 1; it <= ITMAX; it++)
|
||
|
|
{
|
||
|
|
eps = 0;
|
||
|
|
#pragma dvm actual(eps)
|
||
|
|
|
||
|
|
#pragma dvm region
|
||
|
|
{
|
||
|
|
/* Parallel loop with base array A */
|
||
|
|
/* calculating maximum in variable eps */
|
||
|
|
#pragma dvm parallel([i][j][k] on A[i][j][k]) reduction(max(eps)), cuda_block(32, 8)
|
||
|
|
for (i = 1; i < L - 1; i++)
|
||
|
|
for (j = 1; j < L - 1; j++)
|
||
|
|
for (k = 1; k < L - 1; k++)
|
||
|
|
{
|
||
|
|
float tmp = fabs(B[i][j][k] - A[i][j][k]);
|
||
|
|
eps = Max(tmp, eps);
|
||
|
|
A[i][j][k] = B[i][j][k];
|
||
|
|
}
|
||
|
|
|
||
|
|
/* Parallel loop with base array B and */
|
||
|
|
/* with prior updating shadow elements of array A */
|
||
|
|
#pragma dvm parallel([i][j][k] on B[i][j][k]) shadow_renew(A), cuda_block(32, 8)
|
||
|
|
for (i = 1; i < L - 1; i++)
|
||
|
|
for (j = 1; j < L - 1; j++)
|
||
|
|
for (k = 1; k < L - 1; k++)
|
||
|
|
B[i][j][k] = (A[i - 1][j][k] + A[i][j - 1][k] + A[i][j][k - 1] + A[i][j][k + 1] + A[i][j + 1][k] + A[i + 1][j][k]) / 6.0f;
|
||
|
|
}
|
||
|
|
|
||
|
|
#pragma dvm get_actual(eps)
|
||
|
|
printf(" IT = %4i EPS = %14.7E\n", it, eps);
|
||
|
|
if (eps < MAXEPS)
|
||
|
|
break;
|
||
|
|
}
|
||
|
|
#ifdef _DVMH
|
||
|
|
dvmh_barrier();
|
||
|
|
endt = dvmh_wtime();
|
||
|
|
#else
|
||
|
|
endt = 0;
|
||
|
|
#endif
|
||
|
|
|
||
|
|
printf(" Jacobi3D Benchmark Completed.\n");
|
||
|
|
printf(" Size = %4d x %4d x %4d\n", L, L, L);
|
||
|
|
printf(" Iterations = %12d\n", ITMAX);
|
||
|
|
printf(" Time in seconds = %12.2lf\n", endt - startt);
|
||
|
|
printf(" Operation type = floating point\n");
|
||
|
|
printf(" Verification = %12s\n", (fabs(eps - 5.058044) < 1e-4 ? "SUCCESSFUL" : "UNSUCCESSFUL"));
|
||
|
|
|
||
|
|
printf(" END OF Jacobi3D Benchmark\n");
|
||
|
|
return 0;
|
||
|
|
}
|