Below question needs a small code to complete this problem (C language). Problem

Question

Below question needs a small code to complete this problem (C language).

Problem 5 (Dot product for multicore machines ): Complete the SPMD (single program

multiple data) program below for computing dot product on multicore machines using the

collective MPI API’s listed below. The steps are: compute size, scatter part of vectors x and y to

other machines using MPI_Scatter, compute dot product in parallel on all machines and collect

partial products using MPI_Gather, add the partial products to come up with a single final product

by the master machine. No neeed to initialize the vectors.

MPI_Scatter(src_lst,m,MPI_INT,dst_lst,m,MPI_INT,MASTER,world);

MPI_Gather(src_buf,number,MPI_INT,dst_buf,number,MPI_INT,MASTER,world);

int main(int argc, char *argv[]) {

int i,n,nprocs,pid,size,prod=0;

int vector_x[NELMS],vector_y[NELMS],partial_prods[MAXPROCS];

MPI_Status status;

MPI_Comm world;

n = atoi(argv[1]); /* number of elements */

MPI_Init(&argc, &argv);

world = MPI_COMM_WORLD;

MPI_Comm_size(MPI_COMM_WORLD, &nprocs);

MPI_Comm_rank(MPI_COMM_WORLD, &pid);

/////////////////////// Feel code here .....

///////////////////////

MPI_Finalize();

}

int dot_product(int s,int e,int *x,int *y){

int i,prod=0;

for (i=s;i

return prod;

}

Explanation / Answer

Answer:

MPI_Scatter(src_lst,m,MPI_INT,dst_lst,m,MPI_INT,MASTER,world);

MPI_Gather(src_buf,number,MPI_INT,dst_buf,number,MPI_INT,MASTER,world);

int main(int argc, char *argv[])

{

int i,n,nprocs,pid,size,prod=0;

int vector_x[NELMS],vector_y[NELMS],partial_prods[MAXPROCS];

MPI_Status status;

MPI_Comm world;

n = atoi(argv[1]); /* number of elements */

MPI_Init(&argc, &argv);

world = MPI_COMM_WORLD;

MPI_Comm_size(MPI_COMM_WORLD, &nprocs);

MPI_Comm_rank(MPI_COMM_WORLD, &pid);

size=n/nprocs;

MPI_Scatter(vector_x,size,MPI_INT,partial_prods,m,MPI_INT,MASTER,world);

MPI_Scatter(vector_y,size,MPI_INT,partial_prods,m,MPI_INT,MASTER,world);

prod=dot_product(1,NELMS,vector_x,vector_y);

MPI_Gather(&prod,1,MPI_INT,partial_prods,1,MPI_INT,MASTER,world);

printf(" DotProduct value is %d",prod);

MPI_Finalize();

}

int dot_product(int s,int e,int *x,int *y)

{

int i,prod=0;

for (i=s;i<=e;i++)

prod=prod+s[i]*y[i];

return prod;

}

Related Questions

Navigate

• Browse (All)
• Browse B
• Subjects

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.