我要做的是创建一个1和0的数组，并将整个数组拆分为4个进程（不管我的测试用例有多少4个）并让每个进程打印其等级，长度为数组和数组本身。但是当我运行代码时，我会得到类似的结果：
Enter length and sequence length
100 4
1101000010110110101110101
1110101001111100011100100
1011100110011100001011010
1000110101001000001000111
Rank: 0 Length: 25
我不确定为什么会这样。出于某种原因，我无法将数据正确地传递给其他进程（我对MPI来说很新）
这是我的代码：
#include "ore_header.h"
int main(int argc, char** argv) {
srand(time(NULL));
int my_rank, p;
int total;
int length;
int seq_length;
void generate_sequence(int *arr, int n);
int subsequence_check(int *arr,int n, int m);
MPI_Init(&argc, &argv);
MPI_Comm_size(MPI_COMM_WORLD, &p);
MPI_Comm_rank(MPI_COMM_WORLD, &my_rank);
MPI_Request reqs[p];
MPI_Status stats[p];
int p_length;
int p_total[p];
int *buf[p];
if (my_rank == 0) {
printf("Enter length and sequence length\n");
scanf("%d %d",&length, &seq_length);
p_length = length / p;
for (int i = 0; i < p; i++) {
buf[i] = (int*)malloc(p_length*sizeof(int));
generate_sequence(buf[i], p_length); //Generates a sequence
MPI_Isend(&buf[i], p_length, MPI_INT, i, 0, MPI_COMM_WORLD, &reqs[i]); //Send each buffer to each sequence
MPI_Wait(&reqs[my_rank], &stats[my_rank]); //Wait for data to be sent
}
}
MPI_Irecv(&buf[my_rank], p_length, MPI_INT, 0, 0, MPI_COMM_WORLD, &reqs[my_rank]); //Receive Data from process 0
MPI_Wait(&reqs[my_rank], &stats[my_rank]); //Wait for communication to end
printf("\nRank: %d Length: %d\n",my_rank,p_length);
for (int i = 0; i < p_length; i++) {
printf("%d",buf[i]);
}
//tot = subsequence_check(buf,length,seq_length);
//printf("\n\nTotal: %d\n",tot);
MPI_Finalize();
return (0);
}
更新
感谢您提供反馈，我已经修复了问题，现在它按预期工作了：
#include "ore_header.h"
int main(int argc, char** argv) {
srand(time(NULL));
int my_rank, p;
int total;
int length;
int flag;
int seq_length;
void generate_sequence(int *arr, int n);
int subsequence_check(int *arr,int n, int m);
MPI_Init(&argc, &argv);
MPI_Comm_size(MPI_COMM_WORLD, &p);
MPI_Comm_rank(MPI_COMM_WORLD, &my_rank);
int p_length;
int *buf[p];
if (my_rank == 0) {
printf("Enter length and sequence length\n");
scanf("%d %d",&length, &seq_length);
p_length = length / p;
for (int i = 0; i < p; i++) {
buf[i] = (int*)malloc(p_length*sizeof(int));
generate_sequence(buf[i], p_length);
MPI_Send(buf[i], p_length, MPI_INT, i, 0, MPI_COMM_WORLD);
printf("Data sent to process %d\n", i);
}
}
MPI_Bcast(&p_length, 1, MPI_INT, 0, MPI_COMM_WORLD);
buf[my_rank] = (int*)malloc(p_length*sizeof(int));
MPI_Status stats[p];
MPI_Request reqs[p];
MPI_Irecv(buf[my_rank], p_length, MPI_INT, 0, 0, MPI_COMM_WORLD, &reqs[my_rank]);
while(1) {
MPI_Test(&reqs[my_rank], &flag, &stats[my_rank]);
if (flag) break;
}
printf("\nData received on process: %d Length: %d\n",my_rank,p_length);
for (int i = 0; i < p_length; i++) {
printf("%d",buf[my_rank][i]);
}
printf("\n");
//tot = subsequence_check(buf,length,seq_length);
//printf("\n\nTotal: %d\n",tot);
MPI_Finalize();
return (0);
}
这是我的输出：
Enter length and sequence length
100 4
0101100011001110110010011
Data sent to process 0
0010011010000001001010111
Data sent to process 1
0011111111101101100000111
Data sent to process 2
0000000100010011100001111
Data sent to process 3
Data received on process: 0 Length: 25
0101100011001110110010011
Data received on process: 1 Length: 25
0010011010000001001010111
Data received on process: 2 Length: 25
0011111111101101100000111
Data received on process: 3 Length: 25
0000000100010011100001111
答案 0 :(得分：0)
MPI_Isend()需要指向缓冲区的指针，&buf[i]是指向缓冲区的指针。因此，您可以尝试MPI_Isend(buf[i],...吗？
此外，请求reqs[i]在MPI_Isend()中设置，但MPI_Wait()仅检查reqs[my_rank]的完成情况。因此，可以建议另一种修改：
MPI_Isend(buf[i], p_length, MPI_INT, i, 0, MPI_COMM_WORLD, &reqs[i]); //Send each buffer to each sequence
MPI_Wait(&reqs[i], &stats[i]); //Wait for data to be sent
在MPI_Wait()之后调用MPI_Isend()似乎等同于调用MPI_Send()，特别是如果返回的MPI_Status未经MPI_Test_cancelled()之类的测试。在目前的情况下，对MPI_Wait()的连续调用很容易被for循环结束时对MPI_Waitall()的单个调用所取代。
编辑：关于0以外的等级p_length中MPI_Irecv()的值@ a3f的问题仍然有效...