portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-06-17 15:27:06 +01:00
Code Releases Activity

Added streaming stores

Browse Source
This commit is contained in:
Peter Boyle
2015-05-05 18:09:28 +01:00
parent b720222d98
commit 249165d1b2
5 changed files with 71 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

29
lib/simd/Grid_vComplexF.h
View File

@ -176,21 +176,36 @@ namespace Grid {
vsplat(ret,a,b);
}
friend inline void vstore(const vComplexF &ret, ComplexF *a){
friend inline void vstore(const vComplexF &ret, ComplexF *a){
#if defined (AVX1)|| defined (AVX2)
_mm256_store_ps((float *)a,ret.v);
_mm256_store_ps((float *)a,ret.v);
#endif
#ifdef SSE4
_mm_store_ps((float *)a,ret.v);
_mm_store_ps((float *)a,ret.v);
#endif
#ifdef AVX512
_mm512_store_ps((float *)a,ret.v);
//Note v has a3 a2 a1 a0
_mm512_store_ps((float *)a,ret.v);
//Note v has a3 a2 a1 a0
#endif
#ifdef QPX
assert(0);
assert(0);
#endif
}
}
friend inline void vstream(vComplexF &out,const vComplexF &in){
#if defined (AVX1)|| defined (AVX2)
_mm256_stream_ps((float *)&out.v,in.v);
#endif
#ifdef SSE4
_mm_stream_ps((float *)&out.v,in.v);
#endif
#ifdef AVX512
_mm512_stream_ps((float *)&out.v,in.v);
//Note v has a3 a2 a1 a0
#endif
#ifdef QPX
assert(0);
#endif
}
friend inline void vprefetch(const vComplexF &v)
{
_mm_prefetch((const char*)&v.v,_MM_HINT_T0);