#include <stdio.h>
#include <unistd.h>
#include <string.h>
#include <math.h>

/* Include polybench common header. */
#include <polybench.h>

/* Include benchmark-specific header. */
/* Default data type is double, default size is 4000. */
#include "trmm.h"

/* Array initialization. */
static void init_array(int ni,
                       DATA_TYPE *alpha,
                       DATA_TYPE POLYBENCH_2D(A, NI, NI, ni, ni),
                       DATA_TYPE POLYBENCH_2D(B, NI, NI, ni, ni))
{
  int i, j;

  *alpha = 32412;
  for (i = 0; i < ni; i++)
    for (j = 0; j < ni; j++)
    {
      A[i][j] = ((DATA_TYPE)i * j) / ni;
      B[i][j] = ((DATA_TYPE)i * j) / ni;
    }
}

/* DCE code. Must scan the entire live-out data.
   Can be used also to check the correctness of the output. */
static void print_array(int ni,
                        DATA_TYPE POLYBENCH_2D(B, NI, NI, ni, ni))
{
  int i, j;

  for (i = 0; i < ni; i++)
    for (j = 0; j < ni; j++)
    {
      fprintf(stderr, DATA_PRINTF_MODIFIER, B[i][j]);
      if ((i * ni + j) % 20 == 0)
        fprintf(stderr, "\n");
    }
  fprintf(stderr, "\n");
}

/* Main computational kernel. The whole function will be timed,
   including the call and return. */
static void kernel_trmm(int ni,
                        DATA_TYPE alpha,
                        DATA_TYPE POLYBENCH_2D(A, NI, NI, ni, ni),
                        DATA_TYPE POLYBENCH_2D(B, NI, NI, ni, ni))
{
  int i, j, k;

  /*  B := alpha*A'*B, A triangular */
  for (i = 1; i < _PB_NI; i++)
    for (j = 0; j < _PB_NI; j++)
      for (k = 0; k < i; k++)
        B[i][j] += alpha * A[i][k] * B[j][k];
}

int main(int argc, char **argv)
{
  /* Retrieve problem size. */
  int ni = NI;

  /* Variable declaration/allocation. */
  DATA_TYPE alpha;
  POLYBENCH_2D_ARRAY_DECL(A, DATA_TYPE, NI, NI, ni, ni);
  POLYBENCH_2D_ARRAY_DECL(B, DATA_TYPE, NI, NI, ni, ni);

  /* Initialize array(s). */
  init_array(ni, &alpha, POLYBENCH_ARRAY(A), POLYBENCH_ARRAY(B));

  /* Start timer. */
  polybench_start_instruments;

  /* Run kernel. */
  kernel_trmm(ni, alpha, POLYBENCH_ARRAY(A), POLYBENCH_ARRAY(B));

  /* Stop and print timer. */
  polybench_stop_instruments;
  polybench_print_instruments;

  /* Prevent dead-code elimination. All live-out data must be printed
     by the function call in argument. */
  polybench_prevent_dce(print_array(ni, POLYBENCH_ARRAY(B)));

  /* Be clean. */
  POLYBENCH_FREE_ARRAY(A);
  POLYBENCH_FREE_ARRAY(B);

  return 0;
}