portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-04-10 06:00:45 +01:00
Code Releases Activity

Generic or GPU ready for benchmark test on GPU

Browse Source
This commit is contained in:
Peter Boyle 2019-06-05 00:13:52 +01:00
parent ba4fd756b9
commit 6379651cdd
1 changed files with 30 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

48
Grid/qcd/action/fermion/implementation/WilsonKernelsGpuImplementation.h
View File

@ -88,6 +88,7 @@ accelerator_inline void get_stencil(StencilEntry * mem, StencilEntry &chip)
#define GPU_COALESCED_STENCIL_LEG_PROJ(Dir,spProj) \
if (SE._is_local) { \
auto in_t = in[SE._offset+s]; \
decltype(chi) tmp; \
if (SE._permute) { \
spProj(tmp, in_t); \
permute(chi, tmp, ptype); \
@ -105,9 +106,16 @@ accelerator_inline void WilsonKernels<Impl>::GpuDhopSiteDag(StencilView &st, Sit
SiteHalfSpinor *buf, int Ls, int s,
int sU, const FermionFieldView &in, FermionFieldView &out)
{
#ifdef __CUDA_ARCH__
typename SiteHalfSpinor::scalar_object chi;
typename SiteHalfSpinor::scalar_object Uchi;
typename SiteSpinor::scalar_object result;
#else
SiteHalfSpinor chi;
SiteHalfSpinor Uchi;
SiteSpinor result;
#endif
typedef typename SiteSpinor::scalar_type scalar_type;
typedef typename SiteSpinor::vector_type vector_type;
constexpr int Nsimd = sizeof(vector_type)/sizeof(scalar_type);
@ -178,9 +186,15 @@ accelerator_inline void WilsonKernels<Impl>::GpuDhopSite(StencilView &st, SiteDo
SiteHalfSpinor *buf, int Ls, int s,
int sU, const FermionFieldView &in, FermionFieldView &out)
{
#ifdef __CUDA_ARCH__
typename SiteHalfSpinor::scalar_object chi;
typename SiteHalfSpinor::scalar_object Uchi;
typename SiteSpinor::scalar_object result;
#else
SiteHalfSpinor chi;
SiteHalfSpinor Uchi;
SiteSpinor result;
#endif
typedef typename SiteSpinor::scalar_type scalar_type;
typedef typename SiteSpinor::vector_type vector_type;
constexpr int Nsimd = sizeof(vector_type)/sizeof(scalar_type);
@ -282,27 +296,25 @@ void WilsonKernels<Impl>::DhopKernel(int Opt,StencilImpl &st, DoubledGaugeField
auto in_v = in.View();
auto out_v = out.View();
auto st_v = st.View();
if ( (Opt == WilsonKernelsStatic::OptGpu) && interior && exterior ) {
const uint64_t nsimd = Simd::Nsimd();
const uint64_t NN = Nsite*Ls*nsimd;
accelerator_loopN( sss, NN, {
uint64_t cur = sss;
// uint64_t lane = cur % nsimd;
cur = cur / nsimd;
uint64_t s = cur%Ls;
// uint64_t sF = cur;
cur = cur / Ls;
uint64_t sU = cur;
#define KERNEL_CALL(A) \
const uint64_t nsimd = Simd::Nsimd(); \
const uint64_t NN = Nsite*Ls*nsimd;\
accelerator_loopN( sss, NN, { \
uint64_t cur = sss; \
cur = cur / nsimd; \
uint64_t s = cur%Ls; \
cur = cur / Ls; \
uint64_t sU = cur;
WilsonKernels<Impl>::GpuDhopSite(st_v,U_v[sU],buf,Ls,s,sU,in_v,out_v);
});
} else {
/*
accelerator_loop( ss, U_v, {
int sU = ss;
int sF = Ls * sU;
WilsonKernels<Impl>::DhopSite(Opt,st_v,U_v,st.CommBuf(),sF,sU,Ls,1,in_v,out_v);
WilsonKernels<Impl>::GenericDhopSite(Opt,st_v,U_v,st.CommBuf(),sF,sU,Ls,1,in_v,out_v);
});
*/
}
}
template <class Impl>
@ -329,11 +341,11 @@ void WilsonKernels<Impl>::DhopKernel(int Opt,StencilImpl &st, DoubledGaugeField
WilsonKernels<Impl>::GpuDhopSiteDag(st_v,U_v[sU],buf,Ls,s,sU,in_v,out_v);
});
} else {
// accelerator_loop( ss, U_v, {
// int sU = ss;
// int sF = Ls * sU;
// WilsonKernels<Impl>::DhopSiteDag(Opt,st,U_v,st.CommBuf(),sF,sU,Ls,1,in_v,out_v);
// });
accelerator_loop( ss, U_v, {
int sU = ss;
int sF = Ls * sU;
WilsonKernels<Impl>::GenericDhopSiteDag(Opt,st,U_v,st.CommBuf(),sF,sU,Ls,1,in_v,out_v);
});
}
}