trying optimize 3mm

polybench-c-4.2.1-beta/linear-algebra/kernels/3mm/3mm_optimized.c

@@ -81,37 +81,41 @@ void kernel_3mm(int ni, int nj, int nk, int nl, int nm,
   int i, j, k;
 
 #pragma scop
+#pragma omp parallel
+{
+
   /* E := A*B */
-  for (i = 0; i < _PB_NI; i++)
+#pragma omp for schedule(static)
+  for (i = 0; i < _PB_NI; i++){
     for (j = 0; j < _PB_NJ; j++)
     	E[i][j] = SCALAR_VAL(0.0);
-
-  for (i = 0; i < _PB_NI; i++)
     for (k = 0; k < _PB_NK; ++k)
       for (j = 0; j < _PB_NJ; j++)
         E[i][j] += A[i][k] * B[k][j];
-  
+  }
 
   /* F := C*D */
-  for (i = 0; i < _PB_NJ; i++)
+
+#pragma omp for schedule(static)
+  for (i = 0; i < _PB_NJ; i++){
     for (j = 0; j < _PB_NL; j++)
     	F[i][j] = SCALAR_VAL(0.0);
-
-  for (i = 0; i < _PB_NJ; i++)
     for (k = 0; k < _PB_NM; ++k)
       for (j = 0; j < _PB_NL; j++)
         F[i][j] += C[i][k] * D[k][j];
+  }
 
   /* G := E*F */
-  for (i = 0; i < _PB_NI; i++)
+
+#pragma omp for schedule(static)
+  for (i = 0; i < _PB_NI; i++){
     for (j = 0; j < _PB_NL; j++)
 	    G[i][j] = SCALAR_VAL(0.0);
-
-  for (i = 0; i < _PB_NI; i++)
     for (k = 0; k < _PB_NJ; ++k)
       for (j = 0; j < _PB_NL; j++)
         G[i][j] += E[i][k] * F[k][j];
-
+  }
+}
 #pragma endscop
 
 }