openmp simd was failed - c

Question

i write a simple test code to use SIMD in openmp 4.0, but no accelerate i got.
#include<stdio.h>
#include<stdlib.h>
#include<time.h>
#define N 40000000
#pragma omp declare simd
double am(double a, double b){
return a*b+a*b+a*b+a*b+a*b;
}
int main(){
int i,j,s;
double * a=(double *)malloc(sizeof(double)*N);
double * b=(double *)malloc(sizeof(double)*N);
double * c=(double *)malloc(sizeof(double)*N);
for(i=1;i<N;i++){
a[i]=0.56;
b[i]=0.66;
}
clock_t t1=clock();
#pragma omp simd
for(i=0;i<N;i++){
c[i] = am(a[i], b[i]);
}
clock_t t2=clock();
printf("%d\n",(t2-t1));
scanf("%d",&s);
free(a);
free(b);
free(c);
return 0;
}
the time with and without “#pragma omp declare simd” & “#pragma omp simd” are all 2900+ on my SNB.
what's more, warning "...\opm2.c(7,30): warning #13401: vector function was emulated" was accured....i really want to know why.
the same problem also accured in the followed code(enabled c99):
#include<stdio.h>
#include<stdlib.h>
#include<time.h>
int main(){
int n=40000000;
int i,j,s;
double *a,*b,*c;
a=(double *)malloc(sizeof(double)*n);
b=(double *)malloc(sizeof(double)*n);
c=(double *)malloc(sizeof(double)*n);
for(i=1;i<n;i++){
a[i]=0.56;
b[i]=0.66;
}
clock_t t1=clock();
#pragma omp simd
for(i=0;i<n;i++){
c[i]=a[i]*b[i]+a[i]*b[i]+a[i]*b[i]-a[i]*b[i]+a[i]*b[i]+a[i]*b[i];
}
clock_t t2=clock();
printf("%d\n",(t2-t1));
scanf("%d",&s);
free(a);
free(b);
free(c);
return 0;
}
i really enabled openmp and the "#pragma omp padallel for" works well.i use intel c/c++ 2015.
if you can find the reason, really thanks, ありがとう, 多谢啦.

Answer 1

Although I don't know why...but on gcc 4.9 when I use -O1/O2/O3... the "#pragma omp simd" work well, and when I use -O0,it don't work. Ps:if you want to use avx rather than sse, don't forget "-mavx".