01 // Include SIMD Header
02 #ifdef __INTEL_COMPILER
03 #  include <fvec.h>   // also contains xmmintrin.h
04 #else
05 #  include <xmmintrin.h>
06 #endif
07
08 typedef union{
09   __m128 m; float f[4];
10 } vector;
11 
12 // Compute scalar product using Intrisics:
13 float scalarproductIntrinsics(float x[], float y[])
14 {
15   __m128 *vecX = (__m128 *)x;
16   __m128 *vecY = (__m128 *)y;
17   vector tmp;
18 
19   tmp.m = _mm_mul_ps(vecX[0],vecY[0]);
20 
21   return tmp.f[0] + tmp.f[1] + tmp.f[2] + tmp.f[3];
22 }
23
24 // Compute scalar product with F32vec4:
25 float scalarproductSSE(float x[], float y[])
26 {
27   F32vec4 *vecX = (F32vec4 *)x;
28   F32vec4 *vecY = (F32vec4 *)y;
29   F32vec4 tmp;
30 
31   tmp = vecX[0] * vecY[0];
32 
33   return tmp[0] + tmp[1] + tmp[2] + tmp[3];
34 }