portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-06-16 06:47:06 +01:00
Code Releases Activity

Global changes to parallel_for structure.

Browse Source 
Move the comms flags to more sensible names
This commit is contained in:
paboyle
2017-02-21 05:24:27 -05:00
parent 3906cd2149
commit 3ae92fa2e6
43 changed files with 271 additions and 513 deletions
Download Patch File Download Diff File Expand all files Collapse all files

55
lib/lattice/Lattice_local.h
View File

@ -34,47 +34,42 @@ Author: Peter Boyle <paboyle@ph.ed.ac.uk>
namespace Grid {
/////////////////////////////////////////////////////
// Non site, reduced locally reduced routines
/////////////////////////////////////////////////////
// localNorm2,
template<class vobj>
/////////////////////////////////////////////////////
// Non site, reduced locally reduced routines
/////////////////////////////////////////////////////
// localNorm2,
template<class vobj>
inline auto localNorm2 (const Lattice<vobj> &rhs)-> Lattice<typename vobj::tensor_reduced>
{
Lattice<typename vobj::tensor_reduced> ret(rhs._grid);
PARALLEL_FOR_LOOP
for(int ss=0;ss<rhs._grid->oSites(); ss++){
ret._odata[ss]=innerProduct(rhs._odata[ss],rhs._odata[ss]);
}
return ret;
parallel_for(int ss=0;ss<rhs._grid->oSites(); ss++){
ret._odata[ss]=innerProduct(rhs._odata[ss],rhs._odata[ss]);
}
return ret;
}
// localInnerProduct
template<class vobj>
// localInnerProduct
template<class vobj>
inline auto localInnerProduct (const Lattice<vobj> &lhs,const Lattice<vobj> &rhs) -> Lattice<typename vobj::tensor_reduced>
{
Lattice<typename vobj::tensor_reduced> ret(rhs._grid);
PARALLEL_FOR_LOOP
for(int ss=0;ss<rhs._grid->oSites(); ss++){
parallel_for(int ss=0;ss<rhs._grid->oSites(); ss++){
ret._odata[ss]=innerProduct(lhs._odata[ss],rhs._odata[ss]);
}
return ret;
}
// outerProduct Scalar x Scalar -> Scalar
// Vector x Vector -> Matrix
template<class ll,class rr>
// outerProduct Scalar x Scalar -> Scalar
// Vector x Vector -> Matrix
template<class ll,class rr>
inline auto outerProduct (const Lattice<ll> &lhs,const Lattice<rr> &rhs) -> Lattice<decltype(outerProduct(lhs._odata[0],rhs._odata[0]))>
{
Lattice<decltype(outerProduct(lhs._odata[0],rhs._odata[0]))> ret(rhs._grid);
PARALLEL_FOR_LOOP
for(int ss=0;ss<rhs._grid->oSites(); ss++){
ret._odata[ss]=outerProduct(lhs._odata[ss],rhs._odata[ss]);
}
return ret;
}
{
Lattice<decltype(outerProduct(lhs._odata[0],rhs._odata[0]))> ret(rhs._grid);
parallel_for(int ss=0;ss<rhs._grid->oSites(); ss++){
ret._odata[ss]=outerProduct(lhs._odata[ss],rhs._odata[ss]);
}
return ret;
}
}
#endif