我是OpenMP技术和并行编程的初学者。所以，我想知道，如果这段代码的并行版本没问题，或者我需要改进或改变一些东西。
序列号：
#include <time.h>
#include <stdio.h>
#include <stdlib.h>
int main (int argc, char *argv[]) {
int N; //size of columns and rows (matrices)
clock_t start_t, end_t;
double total_t;
time_t time_start, time_stop;
int tid, nthreads, i, j, k, l;
double **a, **b, **c, **d, Z;
double *a_block, *b_block, *c_block, *d_block;
if(argc<2){
printf("Usage: mm matrix_size\n");
exit(-1);
}
N = atoi(argv[1]);
a = (double **) malloc(N*sizeof(double *)); /* matrix a to be multiplied */
b = (double **) malloc(N*sizeof(double *)); /* matrix b to be multiplied */
c = (double **) malloc(N*sizeof(double *)); /* result matrix c */
d = (double **) malloc(N*sizeof(double *)); /* result matrix d */
a_block = (double *) malloc(N*N*sizeof(double)); /* Storage for matrices */
b_block = (double *) malloc(N*N*sizeof(double));
c_block = (double *) malloc(N*N*sizeof(double));
d_block = (double *) malloc(N*N*sizeof(double));
for (i=0; i<N; i++) /* Initialize pointers to a */
a[i] = a_block+i*N;
for (i=0; i<N; i++) /* Initialize pointers to b */
b[i] = b_block+i*N;
for (i=0; i<N; i++) /* Initialize pointers to c */
c[i] = c_block+i*N;
for (i=0; i<N; i++) /* Initialize pointers to d */
d[i] = d_block+i*N;
for (i=0; i<N; i++) /* last matrix has been initialized */
for (j=0; j<N; j++)
a[i][j]= (i+j) * ((double) rand());
for (i=0; i<N; i++)
for (j=0; j<N; j++)
b[i][j]= i * j * ((double) rand());
for (i=0; i<N; i++){
for (j=0; j<N; j++){
c[i][j]= 0.0;
d[i][j]= 0.0;
}
}
/* STEP 1: Z is the (integer) sum of ell elements of a anb b */
start_t = clock();
time_start= time(NULL);
Z=0.0;
/* last matrix has been initialized */
for (i=0; i<N; i++)
for (j=0; j<N; j++)
Z += a[i][j];
for (i=0; i<N; i++)
for (j=0; j<N; j++)
Z += b[i][j];
printf("Value of Z: %e\n", Z);
/* STEP 2: variation of matrix multiplication */
for (i=0; i<N; i++) {
for(j=0; j<N; j++) {
for (k=0; k<N; k++) {
c[i][j] += a[i][k] * b[k][j];
for(l=0; l<i+N+N; l++) {
c[i][j] -= ((Z*k)+l-k)/(l+k);
}
}
}
}
/* STEP 3: add elements of matrices d=a+b */
for (i=0; i<N; i++) {
for(j=0; j<N; j++) {
d[i][j] = a[i][j] + b[i][j];
}
}
end_t = clock();
time_stop = time(NULL);
total_t = (double)(end_t - start_t) / CLOCKS_PER_SEC *1000.0;
printf("Total time taken by CPU: %0.2f seconds. \n", total_t );
printf("Task time is about %0.0f seconds for matrix size equals to %d \n",difftime(time_stop,time_start),N);
printf ("Done.\n");
exit(0);
}
而且，我用这种方式编写了并行版本：
并行代码：
#include <time.h>
#include <omp.h>
#include <stdio.h>
#include <stdlib.h>
int main (int argc, char *argv[]) {
int N; //size of columns and rows (matrices)
clock_t start_t, end_t;
double total_t;
time_t time_start, time_stop;
int tid, nthreads, i, j, k, l;
double **a, **b, **c, **d, Z;
double *a_block, *b_block, *c_block, *d_block;
if(argc<2){
printf("Usage: mm matrix_size\n");
exit(-1);
}
N = atoi(argv[1]);
a = (double **) malloc(N*sizeof(double *)); /* matrix a to be multiplied */
b = (double **) malloc(N*sizeof(double *)); /* matrix b to be multiplied */
c = (double **) malloc(N*sizeof(double *)); /* result matrix c */
d = (double **) malloc(N*sizeof(double *)); /* result matrix d */
a_block = (double *) malloc(N*N*sizeof(double)); /* Storage for matrices */
b_block = (double *) malloc(N*N*sizeof(double));
c_block = (double *) malloc(N*N*sizeof(double));
d_block = (double *) malloc(N*N*sizeof(double));
for (i=0; i<N; i++) /* Initialize pointers to a */
a[i] = a_block+i*N;
for (i=0; i<N; i++) /* Initialize pointers to b */
b[i] = b_block+i*N;
for (i=0; i<N; i++) /* Initialize pointers to c */
c[i] = c_block+i*N;
for (i=0; i<N; i++) /* Initialize pointers to d */
d[i] = d_block+i*N;
for (i=0; i<N; i++) /* last matrix has been initialized */
for (j=0; j<N; j++)
a[i][j]= (i+j) * ((double) rand());
for (i=0; i<N; i++)
for (j=0; j<N; j++)
b[i][j]= i * j * ((double) rand());
for (i=0; i<N; i++){
for (j=0; j<N; j++){
c[i][j]= 0.0;
d[i][j]= 0.0;
}
}
/* STEP 1: Z is the (integer) sum of ell elements of a anb b */
start_t = clock();
time_start= time(NULL);
Z=0.0;
/* last matrix has been initialized */
#pragma omp parallel for private(i,j) reduction(+:Z) schedule(runtime)
for (i=0; i<N; i++)
{
for (j=0; j<N; j++)
{
Z += (a[i][j] + b[i][j]) ;
}
}
printf("Value of Z: %e\n", Z);
#pragma omp parallel sections
{
nthreads = omp_get_num_threads();
/* STEP 2: variation of matrix multiplication */
#pragma omp section
{
#pragma parallel for private(i,j,k,l) schedule(runtime)
for (i=0; i<N; i++) {
for(j=0; j<N; j++) {
for (k=0; k<N; k++) {
c[i][j] += a[i][k] * b[k][j];
for(l=0; l<i+N+N; l++) {
c[i][j] -= ((Z*k)+l-k)/(l+k);
}
}
}
}
}
/* STEP 3: add elements of matrices d=a+b */
#pragma omp section
{
#pragma omp parallel for private(i,j) schedule(runtime)
for (i=0; i<N; i++) {
for(j=0; j<N; j++) {
d[i][j] = a[i][j] + b[i][j];
}
}
}
}
end_t = clock();
time_stop = time(NULL);
total_t = (double)(end_t - start_t) / CLOCKS_PER_SEC *1000.0;
printf("Threads number is : %d. \n",nthreads);
printf("Total time taken by CPU: %0.2f seconds. \n", total_t );
printf("Task time is about %0.0f seconds for matrix size equals to %d \n",difftime(time_stop,time_start),N);
printf ("Done.\n");
exit(0);
}
注意：我使用了运行时调度策略，因为我需要通过OMP_SCHEDULE环境变量进行自定义。
提前致谢！