LAM: please any body solve my problem

/***********  MULTIPLY_MATRICES WITH MPI AND
 CANNON ALGORITHM    *******/

#include <stdio.h> &nbsp; &nbsp; 

#include <math.h>

#include "mpi.h"       

#define N &nbsp; &nbsp; 1200 &nbsp; &nbsp;  /* < 900 ok above 900 problem exists */

#define _mat(i,j) (mat[(i)*n+(j)

 ;

 readmat(char *fname,int *mat,int n)

{

    FILE *fp;

    int i,j;

   
 if((fp=fopen(fname,"r"))==NULL)

      return(-1);

 &nbsp;  for(i=0;i<n;i++)

 &nbsp; &nbsp; &nbsp; for(j=0;j<n;j++)

          if(fscanf(fp,"%d",&_mat(i,j))==EOF)

          {

             fclose(fp);

             return(-1);

          }

    fclose(fp);

    return(0);

}         

writemat(char *fname,int *mat,int n)

{

    FILE *fp;

    int i,j;

    if((fp=fopen(fname,"w"))==NULL)

        return(-1);

 &nbsp;  for(i=0;i<n; fprintf(fp,"n"),i++)

 &nbsp; &nbsp; &nbsp;  for(j=0;j<n;j++)

           fprintf(fp,"%dt",_mat(i,j));

    fclose(fp);

    return(0);

}  

            

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

{

   int
 myrank_old,myrank_new,process_rank,numprocs,situation=0,namelen,source,count;

   double t1,t2;

   

   int M,S,start_block_x,start_block_y;

   int i,j,k,l,coord_nbr_proc;

   float sum;

   int *package_A,*package_B,*C_blocks,*temp_A,*temp_B,*temp_C;

   char processor_name[MPI_MAX_PROCESSOR_NAME]; 

   MPI_Status stat;

   MPI_Comm comm_new;

   int
 num_dims=2,dims[2],periods[2],coords[2];

   int rank_source,rank_dest;

   

   MPI_Init(&;argc,&argv);

   MPI_Comm_rank(MPI_COMM_WORLD,&myrank_old);

   MPI_Comm_size(MPI_COMM_WORLD,&numprocs);

   M=(int)sqrt((double)numprocs);

   S=N/M;

   

   dims[0]=dims[1]=M;

   periods[0]=periods[1]=1;

  
 MPI_Cart_create(MPI_COMM_WORLD,num_dims,dims,periods,0,&comm_new);

   MPI_Comm_rank(comm_new,&myrank_new);

   

   MPI_Get_processor_name(processor_name,&namelen);

   fprintf(stdout,"Process %d of %d on %sn",myrank_new, numprocs, processor_name);

   

   package_A=(int *)malloc(S*S*sizeof(int));

   package_B=(int *)malloc(S*S*sizeof(int));

   C_blocks=(int *)malloc(S*S*sizeof(int));

 

   temp_A=(int
 *)malloc(S*S*sizeof(int));

   temp_B=(int *)malloc(S*S*sizeof(int));

   temp_C=(int *)malloc(S*S*sizeof(int));

       

   

 &nbsp; for(i=0; i<S*S; i++)

        temp_C[i]=0;

   if(myrank_new==0)

   {

       

       int A[N][N],B[N][N],C[N][N];

       

 &nbsp; &nbsp; &nbsp; if(readmat("A_file",(int *)A,N)<0)

           ; situation=1;

          

 &nbsp; &nbsp; &nbsp; if(readmat("B_file",(int *)B,N)<0)

           ; situation=1;

           ; 

       MPI_Bcast(&situation,1,MPI_INT,0,comm_new);

       if(situation==1)

       {   

            ;printf("File A_file or B_file has problem.n");

              MPI_Finalize();

           ; return(0);  

       }

      
 t1=MPI_Wtime();

       

       for(process_rank=1;process_rank<numprocs;process_rank++)

       {

              start_block_x=(process_rank/M);

              start_block_y=(process_rank%M);

           ; 

           ; count=0;

 &nbsp; &nbsp; &nbsp; &nbsp;   ; for(i=start_block_x*S;i<(start_block_x+1)*S;i++)

           ; {

 &nbsp; &nbsp; &nbsp; &nbsp;   ; &nbsp; &nbsp; for(j=start_block_y*S;j<(start_block_y+1)*S;j++)

           ;     {

           ;          package_A[count]=A[i][j];

           ;          package_B[count]=B[i][j];

           ;          count++;

           ;     }

           ;  }

             

           ;  MPI_Send(package_A,S*S,MPI_INT,process_rank,0,comm_new);

           ;  MPI_Send(package_B,S*S,MPI_INT,process_rank,0,comm_new);

        }

        

        start_block_x=(myrank_new/M);

        start_block_y=(myrank_new%M);

        count=0;

        

 &nbsp; &nbsp; &nbsp;  for(i=start_block_x*S;i<(start_block_x+1)*S;i++)

        {

 &nbsp; &nbsp; &nbsp; &nbsp;   ; &nbsp; for(j=start_block_y*S;j<(start_block_y+1)*S;j++)

           ;   {

           ;         temp_A[count]=A[i][j];

           ;         temp_B[count]=B[i][j];

           ;     

           ;         count++;

           ;   }

        }

                    

 &nbsp; &nbsp; &nbsp; &nbsp;for(coord_nbr_proc=0;coord_nbr_proc<;M;coord_nbr_proc++)

        {

           
  

 &nbsp; &nbsp; &nbsp; &nbsp;   ; for(i=0;i<S;i++)

           ; {  

 &nbsp; &nbsp; &nbsp; &nbsp;   ; &nbsp; &nbsp; &nbsp;for(j=0;j<S;j++)

           ;  {         

 &nbsp; &nbsp; &nbsp;   &nbsp; &nbsp; &nbsp; &nbsp;   ; &nbsp; &nbsp; for(k=0;k<;S;k++)

           ;            temp_C[i*S+j]+=temp_A[i*S+k]*temp_B[k*S+j];

                 }       

                   ;

              }

           ;   MPI_Cart_shift(comm_new,1,-1,&rank_source,&rank_dest);

           ;   MPI_Sendrecv_replace(temp_A,S*S,MPI_INT,rank_dest,0,rank_source,0,comm_new,&stat);

           ;   

              MPI_Cart_shift(comm_new,0,-1,&rank_source,&;rank_dest);

           ;  
 MPI_Sendrecv_replace(temp_B,S*S,MPI_INT,rank_dest,0,rank_source,0,comm_new,&stat);

          } 

 &nbsp; &nbsp; &nbsp; &nbsp;  for(i=0;i<S;i++)

          {

 &nbsp; &nbsp; &nbsp; &nbsp;   ;  for(j=0;j<S;j++)

           ;     C[i][j]=temp_C[i*S+j];

          }

        

        

 &nbsp; &nbsp; &nbsp;  for(i=1;i<numprocs;i++)

          {

                   ; 

           ;       

           ;    MPI_Recv(C_blocks,S*S,MPI_INT,MPI_ANY_SOURCE,MPI_ANY_TAG,comm_new,&stat);

           ;        count=0;

           ;    l=0;

           ;    source=stat.MPI_SOURCE;

 &nbsp; &nbsp; &nbsp; &nbsp;   ; &nbsp; &nbsp; &nbsp;  for(j=0;j<S;j++)

           ;   
 {

 &nbsp; &nbsp; &nbsp; &nbsp;   ; &nbsp; &nbsp; &nbsp; &nbsp;   ; for(k=0;k<S;k++)

           ;            {

           ;            ;   C[(source/M)*S+j][(source%M)*S+k]=C_blocks[l*S+count];

           ;            ;   count++;

           ;        } 

           ;        count=0; 

           ;        l++; 

           ;    }      

           ;         

           ;}

         

         t2= MPI_Wtime(); 

           printf("Execution_time=%lfn",(t2-t1));   

         writemat("C_Cannon",(int *)C,N); 

     }          
 

     if(myrank_new!=0)

     {

          

          MPI_Cart_coords(comm_new,myrank_new,2,coords);

          MPI_Bcast(&situation,1,MPI_INT,0,comm_new);    

          if(situation!=0)

          {

           ;  MPI_Finalize();

           ;  return(0);

          }    

          MPI_Recv(temp_A,S*S,MPI_INT,0,0,comm_new,&stat);

          MPI_Recv(temp_B,S*S,MPI_INT,0,0,comm_new,&stat);

         
 MPI_Cart_shift(comm_new,1,-coords[0],&rank_source,&rank_dest);

          MPI_Sendrecv_replace(temp_A,S*S,MPI_INT,rank_dest,0,rank_source,0,comm_new,&stat);

          MPI_Cart_shift(comm_new,0,-coords[1],&rank_source,&rank_dest);

          MPI_Sendrecv_replace(temp_B,S*S,MPI_INT,rank_dest,0,rank_source,0,comm_new,&stat);

 &nbsp; &nbsp;   &nbsp; &nbsp;for(coord_nbr_proc=0;coord_nbr_proc<M;coord_nbr_proc++)

          {

             

 &nbsp; &nbsp; &nbsp; &nbsp;   ; &nbsp; for(i=0;i<S;i++)

           ;   {  

 &nbsp; &nbsp; &nbsp; &nbsp;   ; &nbsp; &nbsp; &nbsp; &nbsp;for(j=0;j<S;j++)

           ;        {

           ;           

 &nbsp; &nbsp; &nbsp;   &nbsp; &nbsp; &nbsp; &nbsp;   ; &nbsp; &nbsp; for(k=0;k<;S;k++)

                   ;          temp_C[i*S+j]+=temp_A[i*S+k]*temp_B[k*S+j];

                   ;     

                   ;  }

              }

           ;   MPI_Cart_shift(comm_new,1,-1,&rank_source,&rank_dest);

           ;   MPI_Sendrecv_replace(temp_A,S*S,MPI_INT,rank_dest,0,rank_source,0,comm_new,&stat);

              MPI_Cart_shift(comm_new,0,-1,&rank_source,&rank_dest);

           ;   MPI_Sendrecv_replace(temp_B,S*S,MPI_INT,rank_dest,0,rank_source,0,comm_new,&stat);

          } 

        MPI_Send(temp_C,S*S,MPI_INT,0,0,comm_new); 

  }    

  MPI_Finalize();

  return(0);           ;  

}