completed P2P blocking call

Author: pradyumnameena

A2/mpi_p2p_b.c

@@ -0,0 +1,190 @@
+#include <mpi.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <math.h>
+#include <time.h>
+
+float tolerance = 0.000001;
+
+float my_random(){
+   float val = (double)rand()/(double)RAND_MAX;
+   return val;
+}
+
+void read_data(float* arr,char str[],int rows,int cols){
+   FILE *file_pointer;
+   file_pointer = fopen(str,"r");
+   for(int i = 0;i<rows;i++){
+      for(int j = 0;j<cols;j++){
+         fscanf(file_pointer,"%f ",&arr[i*cols + j]);
+      }
+   }
+
+   fclose(file_pointer);
+}
+
+void write_data(char file_name[],float* matrix,int rows,int cols){
+   FILE *file_pointer;
+   file_pointer = fopen(file_name,"w");
+   for(int i = 0;i<rows;i++){
+      for(int j = 0;j<cols;j++){
+         fprintf(file_pointer, "%f",matrix[i*cols + j]);
+         if(j!=cols-1){
+            fprintf(file_pointer, " ");
+         }else if(i!=rows-1){
+            fprintf(file_pointer, "\n");
+         }
+      }
+   }
+   fclose(file_pointer);
+   return;
+}
+
+void generate_data(float* arr,int rows,int cols){
+   for(int i = 0;i<rows;i++){
+      for(int j = 0;j<cols;j++){
+         arr[i*cols + j] = my_random();
+      }
+   }
+}
+
+void print(float* array,int rows,int cols){
+   for(int i = 0;i<rows;i++){
+      for(int j = 0;j<cols;j++){
+         printf("%f ",array[i*cols + j]);
+      }
+      printf("\n");
+   }
+   printf("**********THE END**********\n");
+}
+
+void transpose(float* array,float* arrayT,int rows,int cols){
+   for(int i = 0;i<cols;i++){
+      for(int j = 0;j<rows;j++){
+         arrayT[i*rows+j] = array[j*cols + i];
+      }
+   }
+}
+
+void multiply(float* A,float* B,float* prod,int n,int m){
+   int i,j,k;
+   float val = 0;
+
+   for(i = 0;i<n;i++){
+      for(j = 0;j<n;j++){
+         val = 0;
+         for(k = 0;k<m;k++){
+            val+=(A[i*m + k] * B[k*n + j]);
+         }
+         prod[i*n + j] = val;
+      }
+   }
+}
+
+void IsEqual(float* A,float* B,int n){
+   for(int i = 0;i<n;i++){
+      if(fabs(A[i] - B[i])>tolerance){
+         printf("Incorrect\n");
+         return;
+      }
+   }
+   printf("Correct\n");
+}
+
+int main(int argc, char* argv[]){
+   int n = atoi(argv[1]);
+   int m = 32;
+   char* file_name1 = argv[2];
+   char* file_name2 = argv[3];
+   int num_workers = atoi(argv[4]);
+
+   float A[n][m];
+   float B[n][m];
+   float C[m][n];
+   float prodM[n][n];
+   float prodS[n][n];
+
+
+   // Initialize the MPI Environment
+   MPI_Init(&argc,&argv);
+   clock_t t;
+   int rank,idx;
+   MPI_Comm_rank(MPI_COMM_WORLD, &rank);
+
+   int load,extra,num_rows,num_cols;
+
+   if(rank==0){
+      // read_data(A,file_name1,n,m);
+      // read_data(C,file_name2,m,n);
+
+      generate_data(A,n,32);
+      generate_data(C,32,n);
+      transpose(C,B,m,n);
+
+      idx = 0;
+      num_workers-=1;
+      load = n/num_workers;
+      t = clock();
+
+      for(int i = 1;i<=num_workers;i++){
+         if(i==num_workers && num_workers>1){
+            load = n%num_workers;
+         }
+         MPI_Send(&n,1,MPI_INT,i,0,MPI_COMM_WORLD);
+         MPI_Send(&m,1,MPI_INT,i,0,MPI_COMM_WORLD);
+         MPI_Send(&load,1,MPI_INT,i,0,MPI_COMM_WORLD);
+         MPI_Send(&A[idx][0],load*m,MPI_FLOAT,i,0,MPI_COMM_WORLD);
+         MPI_Send(&B[0][0],n*m,MPI_FLOAT,i,0,MPI_COMM_WORLD);
+         idx+=load;
+      }
+
+      idx = 0;
+      load = n/num_workers;
+      for(int i = 1;i<=num_workers;i++){
+         if(i==num_workers && num_workers>1){
+            load = n%num_workers;
+         }
+         MPI_Recv(&prodM[idx][0],load*n,MPI_FLOAT,i,0,MPI_COMM_WORLD,MPI_STATUS_IGNORE);
+         idx+=load;
+      }
+
+      t = clock()-t;
+      double time_taken = ((double)t)/CLOCKS_PER_SEC;
+      printf("Time taken:- %lf\n",time_taken);
+
+      multiply(A,C,prodS,n,m);
+
+      print(prodM,n,n);
+      print(prodS,n,n);
+      IsEqual(prodS,prodM,n*n);
+
+      char file_name[30] = "product_P2PB_";
+      sprintf(file_name+13,"%d",n);
+      strcat(file_name,".txt");
+      write_data(file_name,prodM,n,n);
+
+   }else{
+      MPI_Recv(&num_rows,1,MPI_INT,0,0,MPI_COMM_WORLD,MPI_STATUS_IGNORE);
+      MPI_Recv(&num_cols,1,MPI_INT,0,0,MPI_COMM_WORLD,MPI_STATUS_IGNORE);
+      MPI_Recv(&load,1,MPI_INT,0,0,MPI_COMM_WORLD,MPI_STATUS_IGNORE);
+      MPI_Recv(&A[0][0],load*num_cols,MPI_FLOAT,0,0,MPI_COMM_WORLD,MPI_STATUS_IGNORE);
+      MPI_Recv(&B[0][0],num_rows*num_cols,MPI_FLOAT,0,0,MPI_COMM_WORLD,MPI_STATUS_IGNORE);
+
+      double val;
+      for(int i = 0;i<load;i++){
+         for(int j = 0;j<num_rows;j++){
+            val = 0;
+            for(int k = 0;k<num_cols;k++){
+               val+=(A[i][k]*B[j][k]);
+            }
+            C[i][j] = val;
+         }
+      }
+
+      MPI_Send(&C[0][0],load*num_rows,MPI_FLOAT,0,0,MPI_COMM_WORLD);
+   }
+
+   MPI_Finalize();
+   return 0;
+}