portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-11-04 05:54:32 +00:00
Code Releases Activity

Asm and threading for many core

Browse Source
This commit is contained in:
paboyle
2015-11-06 03:47:14 -08:00
parent b362f8d27b
commit a2ff068e29
1 changed files with 79 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

83
lib/qcd/action/fermion/WilsonFermion5D.cc
View File

@@ -1,4 +1,5 @@
#include <Grid.h>
#include <PerfCount.h>
namespace Grid {
namespace QCD {
@@ -7,6 +8,7 @@ namespace QCD {
const std::vector<int> WilsonFermion5DStatic::directions ({1,2,3,4, 1, 2, 3, 4});
const std::vector<int> WilsonFermion5DStatic::displacements({1,1,1,1,-1,-1,-1,-1});
int WilsonFermion5DStatic::HandOptDslash;
int WilsonFermion5DStatic::AsmOptDslash;
// 5d lattice for DWF.
template<class Impl>
@@ -220,6 +222,13 @@ void WilsonFermion5D<Impl>::DhopInternal(CartesianStencil & st, LebesgueOrder &l
Compressor compressor(dag);
// Assume balanced KMP_AFFINITY; this is forced in GridThread.h
int threads = GridThread::GetThreads();
int HT = GridThread::GetHyperThreads();
int cores = GridThread::GetCores();
int nwork = U._grid->oSites();
st.HaloExchange<SiteSpinor,SiteHalfSpinor,Compressor>(in,comm_buf,compressor);
// Dhop takes the 4d grid from U, and makes a 5d index for fermion
@@ -229,10 +238,13 @@ void WilsonFermion5D<Impl>::DhopInternal(CartesianStencil & st, LebesgueOrder &l
// - 8 linear access unit stride streams per thread for Fermion for hw prefetchable.
if ( dag == DaggerYes ) {
if( this->HandOptDslash ) {
PARALLEL_FOR_LOOP
#pragma omp parallel for schedule(static)
for(int ss=0;ss<U._grid->oSites();ss++){
for(int s=0;s<Ls;s++){
int sU=ss;
if ( LebesgueOrder::UseLebesgueOrder ) {
sU=lo.Reorder(ss);
}
int sF = s+Ls*sU;
Kernels::DiracOptHandDhopSiteDag(st,U,comm_buf,sF,sU,in,out);
}
@@ -251,16 +263,79 @@ PARALLEL_FOR_LOOP
}
}
} else {
if( this->HandOptDslash ) {
PARALLEL_FOR_LOOP
if( this->AsmOptDslash ) {
// for(int i=0;i<1;i++){
// for(int i=0;i< PerformanceCounter::NumTypes(); i++ ){
// PerformanceCounter Counter(i);
// Counter.Start();
#pragma omp parallel for
for(int t=0;t<threads;t++){
int hyperthread = t%HT;
int core = t/HT;
int sswork, swork,soff,ssoff, sU,sF;
GridThread::GetWork(nwork,core,sswork,ssoff,cores);
GridThread::GetWork(Ls , hyperthread, swork, soff,HT);
for(int ss=0;ss<sswork;ss++){
for(int s=soff;s<soff+swork;s++){
sU=ss+ ssoff;
if ( LebesgueOrder::UseLebesgueOrder ) {
sU = lo.Reorder(sU);
}
sF = s+Ls*sU;
Kernels::DiracOptAsmDhopSite(st,U,comm_buf,sF,sU,in,out,(uint64_t *)0);// &buf[0]
}
}
}
// Counter.Stop();
// Counter.Report();
// }
} else if( this->HandOptDslash ) {
#pragma omp parallel for
for(int t=0;t<threads;t++){
int hyperthread = t%HT;
int core = t/HT;
int sswork, swork,soff, sU,sF;
sswork = (nwork + cores-1)/cores;
GridThread::GetWork(Ls , hyperthread, swork, soff,HT);
for(int ss=0;ss<sswork;ss++){
sU=ss+core*sswork; // max locality within an L2 slice
if ( LebesgueOrder::UseLebesgueOrder ) {
sU = lo.Reorder(sU);
}
if ( sU < nwork ) {
for(int s=soff;s<soff+swork;s++){
sF = s+Ls*sU;
Kernels::DiracOptHandDhopSite(st,U,comm_buf,sF,sU,in,out);
}
}
}
}
/*
#pragma omp parallel for schedule(static)
for(int ss=0;ss<U._grid->oSites();ss++){
for(int s=0;s<Ls;s++){
// int sU=lo.Reorder(ss);
int sU=ss;
if ( LebesgueOrder::UseLebesgueOrder ) {
sU=lo.Reorder(ss);
}
int sF = s+Ls*sU;
Kernels::DiracOptHandDhopSite(st,U,comm_buf,sF,sU,in,out);
}
}
*/
} else {
PARALLEL_FOR_LOOP