portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2024-11-15 02:05:37 +00:00
Code Releases Activity

Working on NVCC

Browse Source
This commit is contained in:
Peter Boyle 2024-10-18 13:58:03 -04:00
parent 11b07b950d
commit 955da582b6
1 changed files with 2 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
Grid/qcd/utils/A2Autils.h
View File

@ -354,7 +354,7 @@ void A2Autils<FImpl>::MesonFieldGPU(TensorType &mat,
int Lblock = mat.dimension(3); int Lblock = mat.dimension(3);
int Rblock = mat.dimension(4); int Rblock = mat.dimension(4);
assert(Lblock % block==0); // assert(Lblock % block==0);
// assert(Rblock % block==0); // assert(Rblock % block==0);
GridBase *grid = lhs_wi[0].Grid(); GridBase *grid = lhs_wi[0].Grid();
@ -387,7 +387,6 @@ void A2Autils<FImpl>::MesonFieldGPU(TensorType &mat,
////////////////////////////////////////// //////////////////////////////////////////
// Should write a SpinOuterColorTrace // Should write a SpinOuterColorTrace
////////////////////////////////////////// //////////////////////////////////////////
std::cout <<GridLogMessage << "accelerator_for for i="<<i<<" j="<<j<<std::endl;
t_afor-=usecond(); t_afor-=usecond();
accelerator_for(ss,grid->oSites(),(size_t)Nsimd,{ accelerator_for(ss,grid->oSites(),(size_t)Nsimd,{
auto left = conjugate(lhs_v(ss)); auto left = conjugate(lhs_v(ss));
@ -401,11 +400,10 @@ void A2Autils<FImpl>::MesonFieldGPU(TensorType &mat,
}} }}
coalescedWrite(SpinMat_v[ss],vv); coalescedWrite(SpinMat_v[ss],vv);
}); });
}// j within block
t_afor+=usecond(); t_afor+=usecond();
}// j within block
// After getting the sitewise product do the mom phase loop // After getting the sitewise product do the mom phase loop
for(int m=0;m<Nmom;m++){ for(int m=0;m<Nmom;m++){
std::cout <<GridLogMessage << "mom "<<m<<" i="<<i<<" jo="<<jo<<std::endl;
t_pha-=usecond(); t_pha-=usecond();
MomSpinMat = SpinMat * mom[m]; MomSpinMat = SpinMat * mom[m];
t_pha+=usecond(); t_pha+=usecond();
@ -415,7 +413,6 @@ void A2Autils<FImpl>::MesonFieldGPU(TensorType &mat,
t_sum+=usecond(); t_sum+=usecond();
t_trace-=usecond(); t_trace-=usecond();
for(int mu=0;mu<Ngamma;mu++){ for(int mu=0;mu<Ngamma;mu++){
std::cout <<GridLogMessage << "trace Gamma "<<mu<<std::endl;
for(int t=0;t<sliced.size();t++){ for(int t=0;t<sliced.size();t++){
for(int j=jo;j<MIN(Rblock,jo+block);j++){ for(int j=jo;j<MIN(Rblock,jo+block);j++){
int jj=j%block; int jj=j%block;