portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2024-11-14 01:35:36 +00:00
Code Releases Activity

More cleanup of Grid_simd.h

Browse Source
This commit is contained in:
neo 2015-05-26 13:54:34 +09:00
parent aff978f60a
commit c04cad92ac
2 changed files with 57 additions and 85 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

40
lib/Grid_simd.h
View File

@ -100,46 +100,6 @@ namespace Grid {
#include <simd/Grid_vector_types.h> #include <simd/Grid_vector_types.h>
namespace Grid { namespace Grid {
// NB: Template the following on "type Complex" and then implement *,+,- for
// ComplexF, ComplexD, RealF, RealD above to
// get full generality of binops with scalars.
inline void mac (vComplexF *__restrict__ y,const ComplexF *__restrict__ a,const vComplexF *__restrict__ x){ *y = (*a)*(*x)+(*y); };
inline void mult(vComplexF *__restrict__ y,const ComplexF *__restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) * (*r); }
inline void sub (vComplexF *__restrict__ y,const ComplexF *__restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) - (*r); }
inline void add (vComplexF *__restrict__ y,const ComplexF *__restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) + (*r); }
inline void mac (vComplexF *__restrict__ y,const vComplexF *__restrict__ a,const ComplexF *__restrict__ x){ *y = (*a)*(*x)+(*y); };
inline void mult(vComplexF *__restrict__ y,const vComplexF *__restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) * (*r); }
inline void sub (vComplexF *__restrict__ y,const vComplexF *__restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) - (*r); }
inline void add (vComplexF *__restrict__ y,const vComplexF *__restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) + (*r); }
inline void mac (vComplexD *__restrict__ y,const ComplexD *__restrict__ a,const vComplexD *__restrict__ x){ *y = (*a)*(*x)+(*y); };
inline void mult(vComplexD *__restrict__ y,const ComplexD *__restrict__ l,const vComplexD *__restrict__ r){ *y = (*l) * (*r); }
inline void sub (vComplexD *__restrict__ y,const ComplexD *__restrict__ l,const vComplexD *__restrict__ r){ *y = (*l) - (*r); }
inline void add (vComplexD *__restrict__ y,const ComplexD *__restrict__ l,const vComplexD *__restrict__ r){ *y = (*l) + (*r); }
inline void mac (vComplexD *__restrict__ y,const vComplexD *__restrict__ a,const ComplexD *__restrict__ x){ *y = (*a)*(*x)+(*y); };
inline void mult(vComplexD *__restrict__ y,const vComplexD *__restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) * (*r); }
inline void sub (vComplexD *__restrict__ y,const vComplexD *__restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) - (*r); }
inline void add (vComplexD *__restrict__ y,const vComplexD *__restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) + (*r); }
inline void mac (vRealF *__restrict__ y,const RealF *__restrict__ a,const vRealF *__restrict__ x){ *y = (*a)*(*x)+(*y); };
inline void mult(vRealF *__restrict__ y,const RealF *__restrict__ l,const vRealF *__restrict__ r){ *y = (*l) * (*r); }
inline void sub (vRealF *__restrict__ y,const RealF *__restrict__ l,const vRealF *__restrict__ r){ *y = (*l) - (*r); }
inline void add (vRealF *__restrict__ y,const RealF *__restrict__ l,const vRealF *__restrict__ r){ *y = (*l) + (*r); }
inline void mac (vRealF *__restrict__ y,const vRealF *__restrict__ a,const RealF *__restrict__ x){ *y = (*a)*(*x)+(*y); };
inline void mult(vRealF *__restrict__ y,const vRealF *__restrict__ l,const RealF *__restrict__ r){ *y = (*l) * (*r); }
inline void sub (vRealF *__restrict__ y,const vRealF *__restrict__ l,const RealF *__restrict__ r){ *y = (*l) - (*r); }
inline void add (vRealF *__restrict__ y,const vRealF *__restrict__ l,const RealF *__restrict__ r){ *y = (*l) + (*r); }
inline void mac (vRealD *__restrict__ y,const RealD *__restrict__ a,const vRealD *__restrict__ x){ *y = (*a)*(*x)+(*y); };
inline void mult(vRealD *__restrict__ y,const RealD *__restrict__ l,const vRealD *__restrict__ r){ *y = (*l) * (*r); }
inline void sub (vRealD *__restrict__ y,const RealD *__restrict__ l,const vRealD *__restrict__ r){ *y = (*l) - (*r); }
inline void add (vRealD *__restrict__ y,const RealD *__restrict__ l,const vRealD *__restrict__ r){ *y = (*l) + (*r); }
inline void mac (vRealD *__restrict__ y,const vRealD *__restrict__ a,const RealD *__restrict__ x){ *y = (*a)*(*x)+(*y); };
inline void mult(vRealD *__restrict__ y,const vRealD *__restrict__ l,const RealD *__restrict__ r){ *y = (*l) * (*r); }
inline void sub (vRealD *__restrict__ y,const vRealD *__restrict__ l,const RealD *__restrict__ r){ *y = (*l) - (*r); }
inline void add (vRealD *__restrict__ y,const vRealD *__restrict__ l,const RealD *__restrict__ r){ *y = (*l) + (*r); }
// Default precision // Default precision
#ifdef GRID_DEFAULT_PRECISION_DOUBLE #ifdef GRID_DEFAULT_PRECISION_DOUBLE
typedef vRealD vReal; typedef vRealD vReal;

14
lib/simd/Grid_vector_types.h
View File

@ -2,7 +2,7 @@
/*! @file Grid_vector_types.h /*! @file Grid_vector_types.h
@brief Defines templated class Grid_simd to deal with inner vector types @brief Defines templated class Grid_simd to deal with inner vector types
*/ */
// Time-stamp: <2015-05-26 13:22:36 neo> // Time-stamp: <2015-05-26 13:44:54 neo>
//--------------------------------------------------------------------------- //---------------------------------------------------------------------------
#ifndef GRID_VECTOR_TYPES #ifndef GRID_VECTOR_TYPES
#define GRID_VECTOR_TYPES #define GRID_VECTOR_TYPES
@ -156,6 +156,18 @@ inline void Gpermute(vsimd &y,const vsimd &b,int perm){
friend inline void sub (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) - (*r); } friend inline void sub (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) - (*r); }
friend inline void add (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) + (*r); } friend inline void add (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) + (*r); }
friend inline void mac (Grid_simd *__restrict__ y,const Scalar_type *__restrict__ a,const Grid_simd *__restrict__ x){ *y = (*a)*(*x)+(*y); };
friend inline void mult(Grid_simd *__restrict__ y,const Scalar_type *__restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) * (*r); }
friend inline void sub (Grid_simd *__restrict__ y,const Scalar_type *__restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) - (*r); }
friend inline void add (Grid_simd *__restrict__ y,const Scalar_type *__restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) + (*r); }
friend inline void mac (Grid_simd *__restrict__ y,const Grid_simd *__restrict__ a,const Scalar_type *__restrict__ x){ *y = (*a)*(*x)+(*y); };
friend inline void mult(Grid_simd *__restrict__ y,const Grid_simd *__restrict__ l,const Scalar_type *__restrict__ r){ *y = (*l) * (*r); }
friend inline void sub (Grid_simd *__restrict__ y,const Grid_simd *__restrict__ l,const Scalar_type *__restrict__ r){ *y = (*l) - (*r); }
friend inline void add (Grid_simd *__restrict__ y,const Grid_simd *__restrict__ l,const Scalar_type *__restrict__ r){ *y = (*l) + (*r); }
//not for integer types... //not for integer types...
template < class S = Scalar_type, NotEnableIf<std::is_integral < S >, int> = 0 > template < class S = Scalar_type, NotEnableIf<std::is_integral < S >, int> = 0 >
friend inline Grid_simd adj(const Grid_simd &in){ return conjugate(in); } friend inline Grid_simd adj(const Grid_simd &in){ return conjugate(in); }