Balakumar R <balakumar85@xxxxxxxxx> writes:
> I am trying to use SSE2 instructions to do double precision operations. I am
> unable to compile in GCC.
>
> This is what i got:
>
> __m128d a = {pv1[0],pv1[1]};
> __m128d b = {pv2[0],pv2[1]};
> __m128d sum1;
>
> sum1 = _mm_add_pd(a, b);
>
> and i get the following error: request for member 'm128d_f64' in 'sum1',
> which is of non-class type 'double __vector__'
This complete test case compiled for me with gcc 4.3.2 using -msse2.
#include <emmintrin.h>
double pv1[2];
double pv2[2];
void fn() {
__m128d a = {pv1[0],pv1[1]};
__m128d b = {pv2[0],pv2[1]};
__m128d sum1;
sum1 = _mm_add_pd(a, b);
}
Ian
