devtools/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2026-04-10 11:34:33 +08:00
Code Issues Packages Projects Releases Wiki Activity

Implement AVX512 vectorization of std::complex<float/double>

Browse Source
This commit is contained in:
Gael Guennebaud
2018-12-06 15:58:06 +01:00
parent 3fba59ea59
commit c53eececb0
8 changed files with 587 additions and 65 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
Eigen/src/Core/arch/SSE/PacketMath.h
View File

@@ -61,20 +61,22 @@ template<> struct is_arithmetic<__m128> { enum { value = true }; };
template<> struct is_arithmetic<__m128i> { enum { value = true }; };
template<> struct is_arithmetic<__m128d> { enum { value = true }; };
#define EIGEN_SSE_SHUFFLE_MASK(p,q,r,s) ((s)<<6|(r)<<4|(q)<<2|(p))
#define vec4f_swizzle1(v,p,q,r,s) \
(_mm_castsi128_ps(_mm_shuffle_epi32( _mm_castps_si128(v), ((s)<<6|(r)<<4|(q)<<2|(p)))))
(_mm_castsi128_ps(_mm_shuffle_epi32( _mm_castps_si128(v), EIGEN_SSE_SHUFFLE_MASK(p,q,r,s))))
#define vec4i_swizzle1(v,p,q,r,s) \
(_mm_shuffle_epi32( v, ((s)<<6|(r)<<4|(q)<<2|(p))))
(_mm_shuffle_epi32( v, EIGEN_SSE_SHUFFLE_MASK(p,q,r,s)))
#define vec2d_swizzle1(v,p,q) \
(_mm_castsi128_pd(_mm_shuffle_epi32( _mm_castpd_si128(v), ((q*2+1)<<6|(q*2)<<4|(p*2+1)<<2|(p*2)))))
(_mm_castsi128_pd(_mm_shuffle_epi32( _mm_castpd_si128(v), EIGEN_SSE_SHUFFLE_MASK(2*p,2*p+1,2*q,2*q+1))))
#define vec4f_swizzle2(a,b,p,q,r,s) \
(_mm_shuffle_ps( (a), (b), ((s)<<6|(r)<<4|(q)<<2|(p))))
(_mm_shuffle_ps( (a), (b), EIGEN_SSE_SHUFFLE_MASK(p,q,r,s)))
#define vec4i_swizzle2(a,b,p,q,r,s) \
(_mm_castps_si128( (_mm_shuffle_ps( _mm_castsi128_ps(a), _mm_castsi128_ps(b), ((s)<<6|(r)<<4|(q)<<2|(p))))))
(_mm_castps_si128( (_mm_shuffle_ps( _mm_castsi128_ps(a), _mm_castsi128_ps(b), EIGEN_SSE_SHUFFLE_MASK(p,q,r,s)))))
#define _EIGEN_DECLARE_CONST_Packet4f(NAME,X) \
const Packet4f p4f_##NAME = pset1<Packet4f>(X)