trying optimize 2mm

polybench-c-4.2.1-beta/linear-algebra/build-and-getbest-run.sh

 #!/bin/bash
+rm $1 > /dev/null 2>&1
 
-gcc -O3 -march=native -fopenmp -I ../utilities  ../utilities/polybench.c $1.c -DPOLYBENCH_TIME -D$2_DATASET -o $1 -lm
-
+if [[ $3 == "yes" ]] ; then
+    gcc -O3 -march=native -fopenmp -I ../utilities  ../utilities/polybench.c $1.c -DPOLYBENCH_TIME -D$2_DATASET -o $1 -lm
+else
+    gcc -O3 -march=native -I ../utilities  ../utilities/polybench.c $1.c -DPOLYBENCH_TIME -D$2_DATASET -o $1 -lm
+fi
 
 for i in {0..999}
 do

polybench-c-4.2.1-beta/linear-algebra/kernels/2mm/2mm_optimized.c

@@ -86,11 +86,12 @@ void kernel_2mm(int ni, int nj, int nk, int nl,
 
 #pragma scop
   /* D := alpha*A*B*C + beta*D */
+#pragma omp parallel
+{
 #pragma omp for schedule(static)
   for (i = 0; i < _PB_NI; i++)
     for (j = 0; j < _PB_NJ; j++)
 	    tmp[i][j] = SCALAR_VAL(0.0);
-
 #pragma omp for schedule(static)
   for (i = 0; i < _PB_NI; i++)
     for (k = 0; k < _PB_NK; ++k)
@@ -101,12 +102,12 @@ void kernel_2mm(int ni, int nj, int nk, int nl,
   for (i = 0; i < _PB_NI; i++)
     for (j = 0; j < _PB_NL; j++)
 	    D[i][j] *= beta;
-
 #pragma omp for schedule(static)
   for (i = 0; i < _PB_NI; i++)
     for (k = 0; k < _PB_NJ; ++k)
       for (j = 0; j < _PB_NL; j++)
 	      D[i][j] += tmp[i][k] * C[k][j];
+}
 #pragma endscop
 
 }

polybench-c-4.2.1-beta/linear-algebra/verify-output.sh

 #!/bin/bash
+rm $1 > /dev/null 2>&1
+rm $2 > /dev/null 2>&1
 
 gcc -O3 -march=native -fopenmp -I ../utilities  ../utilities/polybench.c $1.c -DPOLYBENCH_DUMP_ARRAYS -D$3_DATASET -o $1 -lm