Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Highlighted
Beginner
67 Views

Help with sparse matrix multiplication in MKL

I'm trying to multiply sparse x dense matrix using MKL. The output matrix appears to be wrong. Is there any issue with initializing matrix A? I'm running the code on DevCloud. Could someone help me with this? 

#include <stdio.h>
#include <stdlib.h>
#include "mkl.h"
#include "mkl_spblas.h"
#include <stdbool.h>

#define CALL_AND_CHECK_STATUS(function, error_message) do { \
          if(function != SPARSE_STATUS_SUCCESS)             \
          {                                                 \
          printf(error_message); fflush(0);                 \
          status = 1;                                       \
          }                                                 \
} while(0)

/* Consider adjusting LOOP_COUNT based on the performance of your computer */
/* to make sure that total run time is at least 1 second */
#define LOOP_COUNT 10

int main()
{
    double *B, *C, *values, *rv;
    int m, n, p, i, r, j, count, number;
    double alpha, beta;
    double s_initial, s_elapsed;
   
    sparse_matrix_t A = NULL;
    sparse_operation_t operation = SPARSE_OPERATION_NON_TRANSPOSE;
    
    struct matrix_descr descrA;
    descrA.type = SPARSE_MATRIX_TYPE_GENERAL;
    
    sparse_layout_t layout = SPARSE_LAYOUT_ROW_MAJOR;
    sparse_index_base_t indexing = SPARSE_INDEX_BASE_ZERO;
    
    MKL_INT *rows; 
    MKL_INT status;
    MKL_INT *col_indx;
    MKL_INT *bi, *ei, *indx;

    m = p = n = 4; 
    printf (" Initializing data for matrix multiplication C=A*B for matrix \n"
            " A(%ix%i) and matrix B(%ix%i)\n\n", m, p, p, n);
    alpha = 1.0; beta = 0.0;

    printf (" Allocating memory for matrices aligned on 64-byte boundary for better \n"
            " performance \n\n");

   
    values = (double *)mkl_malloc( m *sizeof( double ), 64 );
    B = (double *)mkl_malloc( p*n *sizeof( double ), 64 );
    C = (double *)mkl_malloc( m*n *sizeof( double ), 64 );
    col_indx = (MKL_INT *)mkl_malloc(m *sizeof(MKL_INT), 64);
    rows = (MKL_INT *)mkl_malloc((m + 1) *sizeof(MKL_INT), 64);

    printf (" Intializing matrix data \n\n");
    
    for( i = 0; i < m; i++ )
          values[i] = (rand() % (20 - 1 + 1)) + 1;
    
    for( i = 0; i < m; i++)
          col_indx[i] = i % 3;
    
        
    rows[0] = 0;
    for( i = 1; i < (m + 1); i++ )
          rows[i] = rows[i - 1] + 2;
    
    for (i = 0; i < (p*n); i++) 
    {
        B[i] = (double)(i+1);
    }

    
    for (i = 0; i < (m*n); i++) 
    {
        C[i] = 0.0;
    }


    printf (" Making the first run of matrix product using Intel(R) MKL to get stable run time measurements \n\n");
    
    mkl_sparse_d_create_csr (&A, indexing, m, p, rows, rows + 1, col_indx, values);
    
    mkl_sparse_d_export_csr (A, &indexing, rows, col_indx, &bi, &ei, &indx, &rv);
    
    printf("\nMatrix A \n");
    for(i = 0; i < p; i++) 
    {
        for(long l = 0; l < p; l++) 
        {
                bool flag = false;
                for(long j = bi[i]; j < ei[i]; j++) 
                {
                        if(indx[j] == l) 
                        {
                                flag = true;
                                printf("%.0lf ", rv[j]);
                                break;
                        }
                }
                if(!flag)
                        printf("%d ",0);

        }       printf("\n");
    }
    
    printf("\n Matrix B \n");
    for (i = 0; i < p; i++)
    {
        printf("\n");
        for (j = 0; j < n; j++)
        printf(" %f ",*( B + i * n + j));
    }
    
    CALL_AND_CHECK_STATUS (mkl_sparse_d_mm(operation, alpha, A, descrA, layout, B, n, p, beta, C, m), "Error after MKL_SPARSE_D_MV, csrC*x  \n");

    printf (" Measuring performance of matrix product using Intel(R) MKL \n\n");
    s_initial = dsecnd();
    
    for (r = 0; r < LOOP_COUNT; r++) 
    {
        CALL_AND_CHECK_STATUS (mkl_sparse_d_mm(operation, alpha, A, descrA, layout, B, n, p, beta, C, m), "Error after MKL_SPARSE_D_MV, csrC*x  \n");
    
    }
    
    s_elapsed = (dsecnd() - s_initial) / LOOP_COUNT;

    printf (" == Matrix multiplication using Intel(R) MKL completed == \n"
            " == at %.5f milliseconds == \n\n", (s_elapsed * 1000));
    
    for (i = 0; i < m; i++)
        {
            printf("\n");
            for (j = 0; j < n; j++)
            printf(" %f ", *(C + i * n + j));
        }
    
    printf ("\n Deallocating memory \n\n");
    mkl_free(A);
    mkl_free(B);
    mkl_free(C);
    
    if (s_elapsed < 0.9/LOOP_COUNT) {
        s_elapsed=1.0/LOOP_COUNT/s_elapsed;
        i=(int)(s_elapsed*LOOP_COUNT)+1;
        printf(" It is highly recommended to define LOOP_COUNT for this example on your \n"
               " computer as %i to have total execution time about 1 second for reliability \n"
               " of measurements\n\n", i);
    }

    printf (" Example completed. \n\n");
    return 0;
    
    
    
}
0 Kudos
1 Reply
Highlighted
Moderator
52 Views

>>The output matrix appears to be wrong. Is there any issue with initializing matrix A? 

How could we check the case? You may try to initialize the input data manually then compute and compare the output results.  

 

0 Kudos