#include <omp.h>
#include <stdio.h>
#include <stdlib.h>

int f1( ) {
   int UB = 1000;
   double **a = malloc(UB*sizeof(double *));
   double **b = malloc(UB*sizeof(double *));
   double **c = malloc(UB*sizeof(double *));
   a[0] = malloc(UB*UB*sizeof(double));
   b[0] = malloc(UB*UB*sizeof(double));
   c[0] = malloc(UB*UB*sizeof(double));
   for (int i = 1; i < UB; i++) {
      a[i] = a[0] + (UB * i);
      b[i] = b[0] + (UB * i);
      c[i] = c[0] + (UB * i);
   }

   // initialize the arrays
   for (int i = 0; i < UB; i++) {
      for (int j = 0; j < UB; j++) {
         a[i][j] = 1.0;
         b[i][j] = 1.0;
         c[i][j] = 0.0;
      }
   }

   // Perform the unoptimized matrix multiply
   for (int i = 0; i < UB; i++) {
      for (int j = 0; j < UB; j++) {
         for (int k = 0; k < UB; k++) {
            c[i][j] += a[i][k] + b[k][j];
         }
      }
   }

   // free allocated storage
   free(a[0]); free(b[0]); free(c[0]);
   free(a); free(b); free(c);
}

int f2( ) {
   int UB = 1000;
   // allocate 2 dimensional arrays for A, B and C
   double **a = malloc(UB * sizeof(double *));
   double **b = malloc(UB * sizeof(double *));
   double **c = malloc(UB * sizeof(double *));
   a[0] = malloc(UB*UB*sizeof(double));
   b[0] = malloc(UB*UB*sizeof(double));
   c[0] = malloc(UB*UB*sizeof(double));
   for (int i = 1; i < UB; i++) {
      a[i] = a[0] + (UB * i);
      b[i] = b[0] + (UB * i);
      c[i] = c[0] + (UB * i);
   }

   // initialize the arrays
   for (int i = 0; i < UB; i++) {
      for (int j = 0; j < UB; j++) {
         a[i][j] = 1.0;
         b[i][j] = 1.0;
         c[i][j] = 0.0;
      }
   }

   for (int i = 0; i < UB; i++) {
      for (int j = 0; j < UB; j++) {
         for (int k = 0; k < UB; k++) {
            c[i][j] += a[i][k] + b[k][j];
         }
      }
   }
   // free allocated storage
   free(a[0]); free(b[0]); free(c[0]);
   free(a); free(b); free(c);
}

int main (int argc, char *argv[]) {
   double t = omp_get_wtime( );
   // make these two calls run in parallel using OpenMP parallel sections
   // run on scholar
   #pragma omp parallel sections
   {
      #pragma omp section
      {
      f1( );
      }
      #pragma omp section
      {
      f2( );
      }
   }
   t = omp_get_wtime( ) - t;
   printf("time for sequential lab3d is %g\n", t);
}