#include<stdio.h>
#include<stdlib.h>
#include "cblas.h"

int main(){

int n = 14000;
int incx = 1;
double *x, ddt;

x = (double*)malloc(n*sizeof(double));

for (int i=0; i<n; i++){
     x[i] = 1.0;
}


#pragma omp parallel
{
#pragma omp parallel for
for (int i=0; i<10; i++){
   ddt = cblas_ddot(n, x, incx, x, incx);
}
}
printf("dot(x, x) = %f", ddt);
free(x);
}