portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-04-11 14:40:46 +01:00
Code Releases Activity

commVector different for SHMEM compat

Browse Source
This commit is contained in:
paboyle 2016-10-20 16:58:53 +01:00
parent 5fe2b85cbd
commit 9b39f35ae6
1 changed files with 5 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
lib/cshift/Cshift_common.h
View File

@ -45,7 +45,7 @@ public:
// Gather for when there is no need to SIMD split with compression // Gather for when there is no need to SIMD split with compression
/////////////////////////////////////////////////////////////////// ///////////////////////////////////////////////////////////////////
template<class vobj,class cobj,class compressor> void template<class vobj,class cobj,class compressor> void
Gather_plane_simple (const Lattice<vobj> &rhs,std::vector<cobj,alignedAllocator<cobj> > &buffer,int dimension,int plane,int cbmask,compressor &compress, int off=0) Gather_plane_simple (const Lattice<vobj> &rhs,commVector<cobj> &buffer,int dimension,int plane,int cbmask,compressor &compress, int off=0)
{ {
int rd = rhs._grid->_rdimensions[dimension]; int rd = rhs._grid->_rdimensions[dimension];
@ -114,6 +114,7 @@ PARALLEL_NESTED_LOOP2
int o = n*n1; int o = n*n1;
int offset = b+n*n2; int offset = b+n*n2;
cobj temp =compress(rhs._odata[so+o+b]); cobj temp =compress(rhs._odata[so+o+b]);
extract<cobj>(temp,pointers,offset); extract<cobj>(temp,pointers,offset);
} }
@ -121,6 +122,7 @@ PARALLEL_NESTED_LOOP2
} else { } else {
assert(0); //Fixme think this is buggy assert(0); //Fixme think this is buggy
for(int n=0;n<e1;n++){ for(int n=0;n<e1;n++){
for(int b=0;b<e2;b++){ for(int b=0;b<e2;b++){
int o=n*rhs._grid->_slice_stride[dimension]; int o=n*rhs._grid->_slice_stride[dimension];
@ -139,7 +141,7 @@ PARALLEL_NESTED_LOOP2
////////////////////////////////////////////////////// //////////////////////////////////////////////////////
// Gather for when there is no need to SIMD split // Gather for when there is no need to SIMD split
////////////////////////////////////////////////////// //////////////////////////////////////////////////////
template<class vobj> void Gather_plane_simple (const Lattice<vobj> &rhs,std::vector<vobj,alignedAllocator<vobj> > &buffer, int dimension,int plane,int cbmask) template<class vobj> void Gather_plane_simple (const Lattice<vobj> &rhs,commVector<vobj> &buffer, int dimension,int plane,int cbmask)
{ {
SimpleCompressor<vobj> dontcompress; SimpleCompressor<vobj> dontcompress;
Gather_plane_simple (rhs,buffer,dimension,plane,cbmask,dontcompress); Gather_plane_simple (rhs,buffer,dimension,plane,cbmask,dontcompress);
@ -157,7 +159,7 @@ template<class vobj> void Gather_plane_extract(const Lattice<vobj> &rhs,std::vec
////////////////////////////////////////////////////// //////////////////////////////////////////////////////
// Scatter for when there is no need to SIMD split // Scatter for when there is no need to SIMD split
////////////////////////////////////////////////////// //////////////////////////////////////////////////////
template<class vobj> void Scatter_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllocator<vobj> > &buffer, int dimension,int plane,int cbmask) template<class vobj> void Scatter_plane_simple (Lattice<vobj> &rhs,commVector<vobj> &buffer, int dimension,int plane,int cbmask)
{ {
int rd = rhs._grid->_rdimensions[dimension]; int rd = rhs._grid->_rdimensions[dimension];