#include <stdio.h>
#include <stdlib.h>
#include <sys/times.h>
#include <unistd.h>
#include <math.h>


#ifndef MINDIM_M
#define MINDIM_M    1000
#endif

#ifndef MINDIM_N
#define MINDIM_N    1000
#endif

#ifndef MINDIM_K
#define MINDIM_K    1000
#endif

#ifndef MAXDIM_M
#define MAXDIM_M    5000
#endif

#ifndef MAXDIM_N
#define MAXDIM_N    5000
#endif

#ifndef MAXDIM_K
#define MAXDIM_K    5000
#endif

#ifndef INC_M
#define INC_M   100
#endif

#ifndef INC_N
#define INC_N   100
#endif

#ifndef INC_K
#define INC_K   100
#endif

#ifndef MIN_T
#define MIN_T   1
#endif

double buffer[MAXDIM_M*MAXDIM_N];
double buffer1[MAXDIM_M*MAXDIM_N];
double buffer2[MAXDIM_M*MAXDIM_N];

double
walltime()
{
   struct tms    ts;
   static double ClockTick=0.0;

   if (ClockTick==0.0) {
        ClockTick = 1.0 / ((double) sysconf(_SC_CLK_TCK));
   }
   return ((double) times(&ts)) * ClockTick;
}

void
gecopy(long m, long n, const double *A, long incRowA, long incColA,
       double *B, long incRowB, long incColB)
{
    long i, j;

    for (i=0; i<m; ++i) {
        for (j=0; j<n; ++j) {
            B[i*incRowB+j*incColB] = A[i*incRowA+j*incColA];
        }
    }
}

double
asumDiff(long m, long n, const double *A, long incRowA, long incColA,
         const double *B, long incRowB, long incColB)
{
    double diff = 0;
    long   i, j;

    for (i=0; i<m; ++i) {
        for (j=0; j<n; ++j) {
            diff += fabs(B[i*incRowB+j*incColB] - A[i*incRowA+j*incColA]);
        }
    }
    return diff;
}


//------------------------------------------------------------------------------

void
initMatrix(long m, long n, double *A, long incRowA, long incColA)
{
    long i, j;

    for (j=0; j<n; ++j) {
        for (i=0; i<m; ++i) {
            A[i*incRowA+j*incColA] = j*n+i+1;
        }
    }
}

void
printMatrix(long m, long n, const double *A, long incRowA, long incColA)
{
    long i, j;

    for (i=0; i<m; ++i) {
        printf("   ");
        for (j=0; j<n; ++j) {
            printf("%4.1lf ", A[i*incRowA+j*incColA]);
        }
        printf("\n");
    }
    printf("\n");
}

int
main()
{
    long    m, n, runs;
    double  t0, t1, t2, diff;

    printf("#  M    N     t1 (colMajor)  t2 (rowMajor)      t2/t1        diff\n");
    printf("#================================================================\n");

    for (m=MINDIM_M, n=MINDIM_N; m<=MAXDIM_M && n<MAXDIM_N; m+=INC_M, n+=INC_N) {

        printf("%4ld %4ld ", m, n);

        runs = 0;
        t1   = 0;
        do {
            t0 = walltime();
            initMatrix(m, n, buffer, 1, m);
            t1 += walltime() - t0;
            ++runs;
        } while (t1<MIN_T);
        t1 /= runs;
        gecopy(m, n, buffer, 1, m, buffer1, 1, m);

        printf(" %12.2lf", t1);

        runs = 0;
        t2 = 0;
        do {
            t0 = walltime();
            initMatrix(m, n, buffer, n, 1);
            t2 += walltime() - t0;
            ++runs;
        } while (t2<MIN_T);
        t2 /= runs;
        gecopy(m, n, buffer, n, 1, buffer2, n, 1);

        diff = asumDiff(m, n, buffer1, 1, m, buffer2, n, 1);

        printf(" %12.2lf %16.2lf", t2, t2/t1);
        printf(" %11.2le", diff);
        printf("\n");
    }

    return 0;
}