portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-06-18 07:47:06 +01:00
Code Releases Activity

Improvements to the assembler interface that let us move chunks of the

Browse Source 
site and s loop into the kernels. This will save on function call overhead and
guarantee L2 prefetching strategy is right since OMP can't distribute the
sub-chunks of work.
This commit is contained in:
paboyle
2016-06-09 01:12:36 -07:00
parent d9408893b3
commit 55f65b81b5
10 changed files with 77 additions and 87 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
lib/qcd/action/fermion/WilsonFermion.cc
View File

@ -296,12 +296,12 @@ PARALLEL_FOR_LOOP
if ( dag == DaggerYes ) {
PARALLEL_FOR_LOOP
for(int sss=0;sss<in._grid->oSites();sss++){
Kernels::DiracOptDhopSiteDag(st,U,st.comm_buf,sss,sss,in,out);
Kernels::DiracOptDhopSiteDag(st,U,st.comm_buf,sss,sss,1,1,in,out);
}
} else {
PARALLEL_FOR_LOOP
for(int sss=0;sss<in._grid->oSites();sss++){
Kernels::DiracOptDhopSite(st,U,st.comm_buf,sss,sss,in,out);
Kernels::DiracOptDhopSite(st,U,st.comm_buf,sss,sss,1,1,in,out);
}
}
};