portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2024-11-09 23:45:36 +00:00
Code Releases Activity

Changes premerge to develop

Browse Source
This commit is contained in:
Peter Boyle 2023-03-28 20:04:21 -07:00
parent 6af97069b9
commit 4a261fab30
2 changed files with 6 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
Grid/lattice/Lattice_rng.h
View File

@ -440,17 +440,8 @@ public:
_grid->GlobalCoorToGlobalIndex(gcoor,gidx);
_grid->GlobalCoorToRankIndex(rank,o_idx,i_idx,gcoor);
#if 1
assert(rank == _grid->ThisRank() );
#else
//
if (rank != _grid->ThisRank() ){
std::cout <<"rank "<<rank<<" _grid->ThisRank() "<<_grid->ThisRank()<< std::endl;
// exit(-42);
// assert(0);
}
#endif
assert(rank == _grid->ThisRank() );
int l_idx=generator_idx(o_idx,i_idx);
_generators[l_idx] = master_engine;

9
Grid/qcd/action/fermion/implementation/WilsonKernelsImplementation.h
View File

@ -462,6 +462,7 @@ void WilsonKernels<Impl>::DhopKernel(int Opt,StencilImpl &st, DoubledGaugeField
autoView(st_v , st,AcceleratorRead);
if( interior && exterior ) {
acceleratorFenceComputeStream();
if (Opt == WilsonKernelsStatic::OptGeneric ) { KERNEL_CALL(GenericDhopSite); return;}
if (Opt == WilsonKernelsStatic::OptHandUnroll ) { KERNEL_CALL(HandDhopSite); return;}
#ifndef GRID_CUDA
@ -474,6 +475,7 @@ void WilsonKernels<Impl>::DhopKernel(int Opt,StencilImpl &st, DoubledGaugeField
if (Opt == WilsonKernelsStatic::OptInlineAsm ) { ASM_CALL(AsmDhopSiteInt); return;}
#endif
} else if( exterior ) {
acceleratorFenceComputeStream();
if (Opt == WilsonKernelsStatic::OptGeneric ) { KERNEL_CALL(GenericDhopSiteExt); return;}
if (Opt == WilsonKernelsStatic::OptHandUnroll ) { KERNEL_CALL(HandDhopSiteExt); return;}
#ifndef GRID_CUDA
@ -493,15 +495,15 @@ void WilsonKernels<Impl>::DhopKernel(int Opt,StencilImpl &st, DoubledGaugeField
autoView(st_v ,st,AcceleratorRead);
if( interior && exterior ) {
acceleratorFenceComputeStream();
if (Opt == WilsonKernelsStatic::OptGeneric ) { KERNEL_CALL(GenericDhopSiteDag); return;}
if (Opt == WilsonKernelsStatic::OptHandUnroll ) { KERNEL_CALL(HandDhopSiteDag); return;}
#ifndef GRID_CUDA
if (Opt == WilsonKernelsStatic::OptInlineAsm ) { ASM_CALL(AsmDhopSiteDag); return;}
#endif
acceleratorFenceComputeStream();
} else if( interior ) {
if (Opt == WilsonKernelsStatic::OptGeneric ) { KERNEL_CALL(GenericDhopSiteDagInt); return;}
if (Opt == WilsonKernelsStatic::OptHandUnroll ) { KERNEL_CALL(HandDhopSiteDagInt); return;}
if (Opt == WilsonKernelsStatic::OptGeneric ) { KERNEL_CALLNB(GenericDhopSiteDagInt); return;}
if (Opt == WilsonKernelsStatic::OptHandUnroll ) { KERNEL_CALLNB(HandDhopSiteDagInt); return;}
#ifndef GRID_CUDA
if (Opt == WilsonKernelsStatic::OptInlineAsm ) { ASM_CALL(AsmDhopSiteDagInt); return;}
#endif
@ -512,7 +514,6 @@ void WilsonKernels<Impl>::DhopKernel(int Opt,StencilImpl &st, DoubledGaugeField
#ifndef GRID_CUDA
if (Opt == WilsonKernelsStatic::OptInlineAsm ) { ASM_CALL(AsmDhopSiteDagExt); return;}
#endif
acceleratorFenceComputeStream();
}
assert(0 && " Kernel optimisation case not covered ");
}