portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-11-03 21:44:33 +00:00
Code Releases Activity

Merge branch 'master' of https://github.com/paboyle/Grid into scidac1_2

Browse Source 
Conflicts:
	lib/qcd/action/fermion/WilsonKernels.h
	tests/Make.inc
This commit is contained in:
Jung
2015-12-15 11:11:59 -05:00
parent 284453c5e9 af855cc129
commit bc34b7e808
36 changed files with 9091 additions and 185 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
lib/qcd/action/fermion/CayleyFermion5D.cc
View File

@@ -114,7 +114,7 @@ namespace QCD {
// Apply Dw
this->DW(psi,Din,DaggerYes);
Meooe5D(Din,chi);
MeooeDag5D(Din,chi);
int Ls=this->Ls;
for(int s=0;s<Ls;s++){
@@ -163,7 +163,6 @@ namespace QCD {
FermionField tmp(psi._grid);
// Assemble the 5d matrix
Meooe5D(psi,tmp);
#if 0
std::cout << "Meooe Test replacement norm2 tmp = " <<norm2(tmp)<<std::endl;
for(int s=0;s<Ls;s++){
@@ -202,7 +201,7 @@ namespace QCD {
this->DhopOE(psi,tmp,DaggerYes);
}
Meooe5D(tmp,chi);
MeooeDag5D(tmp,chi);
#if 0
std::cout << "Meooe Test replacement norm2 chi new = " <<norm2(chi)<<std::endl;
// Assemble the 5d matrix

21
lib/qcd/action/fermion/WilsonKernels.cc
View File

@@ -3,9 +3,12 @@ namespace Grid {
namespace QCD {
template<class Impl>
void WilsonKernels<Impl>::DiracOptDhopSite(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int sF,int sU,const FermionField &in, FermionField &out)
WilsonKernels<Impl>::WilsonKernels(const ImplParams &p): Base(p) {};
template<class Impl>
void WilsonKernels<Impl>::DiracOptDhopSiteDag(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int sF,int sU,const FermionField &in, FermionField &out)
{
SiteHalfSpinor tmp;
SiteHalfSpinor chi;
@@ -122,7 +125,7 @@ void WilsonKernels<Impl>::DiracOptDhopSite(StencilImpl &st,DoubledGaugeField &U,
};
template<class Impl>
void WilsonKernels<Impl>::DiracOptDhopSiteDag(StencilImpl &st,DoubledGaugeField &U,
void WilsonKernels<Impl>::DiracOptDhopSite(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int sF,int sU,const FermionField &in, FermionField &out)
{
@@ -369,6 +372,16 @@ void WilsonKernels<Impl>::DiracOptDhopDir(StencilImpl &st,DoubledGaugeField &U,
vstream(out._odata[sF],result*(-0.5));
}
#if ( ! defined(AVX512) ) && ( ! defined(IMCI) )
template<class Impl>
void WilsonKernels<Impl>::DiracOptAsmDhopSite(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int sF,int sU,const FermionField &in, FermionField &out,uint64_t *p)
{
DiracOptDhopSite(st,U,buf,sF,sU,in,out); // will template override for Wilson Nc=3
}
#endif
FermOpTemplateInstantiate(WilsonKernels);
GparityFermOpTemplateInstantiate(WilsonKernels);

26
lib/qcd/action/fermion/WilsonKernels.h
View File

@@ -28,11 +28,11 @@ namespace Grid {
void DiracOptDhopDir(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int sF,int sU,const FermionField &in, FermionField &out,int dirdisp,int gamma);
#if defined(AVX512) || defined(IMCI)
void DiracOptAsmDhopSite(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int sF,int sU,const FermionField &in, FermionField &out,uint64_t *);
#else
#if 0
void DiracOptAsmDhopSite(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int sF,int sU,const FermionField &in, FermionField &out,uint64_t *p){
@@ -41,7 +41,7 @@ namespace Grid {
#endif
// doesn't seem to work with Gparity at the moment
#undef HANDOPT
#ifdef HANDOPT
//#define HANDOPT
void DiracOptHandDhopSite(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int sF,int sU,const FermionField &in, FermionField &out);
@@ -49,25 +49,9 @@ namespace Grid {
void DiracOptHandDhopSiteDag(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int sF,int sU,const FermionField &in, FermionField &out);
#else
void DiracOptHandDhopSite(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int sF,int sU,const FermionField &in, FermionField &out)
{
DiracOptDhopSite(st,U,buf,sF,sU,in,out); // will template override for Wilson Nc=3
}
void DiracOptHandDhopSiteDag(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int sF,int sU,const FermionField &in, FermionField &out)
{
DiracOptDhopSiteDag(st,U,buf,sF,sU,in,out); // will template override for Wilson Nc=3
}
#endif
WilsonKernels(const ImplParams &p= ImplParams()) : Base(p) {};
WilsonKernels(const ImplParams &p= ImplParams());
};
}

36
lib/qcd/action/fermion/WilsonKernelsHand.cc
View File

@@ -282,7 +282,7 @@ namespace QCD {
#ifdef HANDOPT
template<class Impl>
void WilsonKernels<Impl >::DiracOptHandDhopSite(StencilImpl &st,DoubledGaugeField &U,
void WilsonKernels<Impl >::DiracOptHandDhopSiteDag(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int ss,int sU,const FermionField &in, FermionField &out)
{
@@ -526,7 +526,7 @@ void WilsonKernels<Impl >::DiracOptHandDhopSite(StencilImpl &st,DoubledGaugeFiel
}
template<class Impl>
void WilsonKernels<Impl >::DiracOptHandDhopSiteDag(StencilImpl &st,DoubledGaugeField &U,
void WilsonKernels<Impl >::DiracOptHandDhopSite(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int ss,int sU,const FermionField &in, FermionField &out)
{
@@ -767,6 +767,36 @@ void WilsonKernels<Impl >::DiracOptHandDhopSiteDag(StencilImpl &st,DoubledGaugeF
vstream(ref()(3)(2),result_32*(-0.5));
}
}
FermOpTemplateInstantiate(WilsonKernels);
#else
template<class Impl>
void WilsonKernels<Impl >::DiracOptHandDhopSite(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int ss,int sU,const FermionField &in, FermionField &out)
{
DiracOptDhopSite(st,U,buf,sF,sU,in,out); // will template override for Wilson Nc=3
}
template<class Impl>
void WilsonKernels<Impl >::DiracOptHandDhopSiteDag(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int ss,int sU,const FermionField &in, FermionField &out)
{
DiracOptDhopSiteDag(st,U,buf,sF,sU,in,out); // will template override for Wilson Nc=3
}
#endif
template void WilsonKernels<WilsonImplF>::DiracOptHandDhopSite(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int ss,int sU,const FermionField &in, FermionField &out);
template void WilsonKernels<WilsonImplD>::DiracOptHandDhopSite(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int ss,int sU,const FermionField &in, FermionField &out);
template void WilsonKernels<WilsonImplF>::DiracOptHandDhopSiteDag(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int ss,int sU,const FermionField &in, FermionField &out);
template void WilsonKernels<WilsonImplD>::DiracOptHandDhopSiteDag(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int ss,int sU,const FermionField &in, FermionField &out);
}}

96
lib/qcd/spin/TwoSpinor.h
View File

@@ -45,13 +45,13 @@ namespace QCD {
// To fail is not to err (Cryptic clue: suggest to Google SFINAE ;) )
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProjXp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
{
hspin(0)=fspin(0)-timesI(fspin(3));
hspin(1)=fspin(1)-timesI(fspin(2));
hspin(0)=fspin(0)+timesI(fspin(3));
hspin(1)=fspin(1)+timesI(fspin(2));
}
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProjXm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
{
hspin(0)=fspin(0)+timesI(fspin(3));
hspin(1)=fspin(1)+timesI(fspin(2));
hspin(0)=fspin(0)-timesI(fspin(3));
hspin(1)=fspin(1)-timesI(fspin(2));
}
// 0 0 0 -1 [0] -+ [3]
@@ -60,13 +60,13 @@ namespace QCD {
// -1 0 0 0
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProjYp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
{
hspin(0)=fspin(0)+fspin(3);
hspin(1)=fspin(1)-fspin(2);
hspin(0)=fspin(0)-fspin(3);
hspin(1)=fspin(1)+fspin(2);
}
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProjYm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
{
hspin(0)=fspin(0)-fspin(3);
hspin(1)=fspin(1)+fspin(2);
hspin(0)=fspin(0)+fspin(3);
hspin(1)=fspin(1)-fspin(2);
}
/*Gz
* 0 0 i 0 [0]+-i[2]
@@ -76,14 +76,14 @@ namespace QCD {
*/
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProjZp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
{
hspin(0)=fspin(0)-timesI(fspin(2));
hspin(1)=fspin(1)+timesI(fspin(3));
hspin(0)=fspin(0)+timesI(fspin(2));
hspin(1)=fspin(1)-timesI(fspin(3));
}
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProjZm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
{
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
hspin(0)=fspin(0)+timesI(fspin(2));
hspin(1)=fspin(1)-timesI(fspin(3));
hspin(0)=fspin(0)-timesI(fspin(2));
hspin(1)=fspin(1)+timesI(fspin(3));
}
/*Gt
* 0 0 1 0 [0]+-[2]
@@ -94,14 +94,14 @@ namespace QCD {
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProjTp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
{
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
hspin(0)=fspin(0)-fspin(2);
hspin(1)=fspin(1)-fspin(3);
hspin(0)=fspin(0)+fspin(2);
hspin(1)=fspin(1)+fspin(3);
}
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProjTm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
{
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
hspin(0)=fspin(0)+fspin(2);
hspin(1)=fspin(1)+fspin(3);
hspin(0)=fspin(0)-fspin(2);
hspin(1)=fspin(1)-fspin(3);
}
/*G5
* 1 0 0 0
@@ -157,32 +157,32 @@ namespace QCD {
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
fspin(0)=hspin(0);
fspin(1)=hspin(1);
fspin(2)=timesI(hspin(1));
fspin(3)=timesI(hspin(0));
fspin(2)=timesMinusI(hspin(1));
fspin(3)=timesMinusI(hspin(0));
}
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spReconXm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
{
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
fspin(0)=hspin(0);
fspin(1)=hspin(1);
fspin(2)=timesMinusI(hspin(1));
fspin(3)=timesMinusI(hspin(0));
fspin(2)=timesI(hspin(1));
fspin(3)=timesI(hspin(0));
}
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumReconXp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
{
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
fspin(0)+=hspin(0);
fspin(1)+=hspin(1);
fspin(2)+=timesI(hspin(1));
fspin(3)+=timesI(hspin(0));
fspin(2)-=timesI(hspin(1));
fspin(3)-=timesI(hspin(0));
}
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumReconXm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
{
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
fspin(0)+=hspin(0);
fspin(1)+=hspin(1);
fspin(2)-=timesI(hspin(1));
fspin(3)-=timesI(hspin(0));
fspin(2)+=timesI(hspin(1));
fspin(3)+=timesI(hspin(0));
}
// 0 0 0 -1 [0] -+ [3]
@@ -195,32 +195,32 @@ namespace QCD {
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
fspin(0)=hspin(0);
fspin(1)=hspin(1);
fspin(2)=-hspin(1);
fspin(3)= hspin(0);
fspin(2)= hspin(1);
fspin(3)=-hspin(0);//Unary minus?
}
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spReconYm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
{
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
fspin(0)=hspin(0);
fspin(1)=hspin(1);
fspin(2)= hspin(1);
fspin(3)=-hspin(0);//Unary minus?
fspin(2)=-hspin(1);
fspin(3)= hspin(0);
}
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumReconYp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
{
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
fspin(0)+=hspin(0);
fspin(1)+=hspin(1);
fspin(2)-=hspin(1);
fspin(3)+=hspin(0);
fspin(2)+=hspin(1);
fspin(3)-=hspin(0);
}
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumReconYm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
{
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
fspin(0)+=hspin(0);
fspin(1)+=hspin(1);
fspin(2)+=hspin(1);
fspin(3)-=hspin(0);
fspin(2)-=hspin(1);
fspin(3)+=hspin(0);
}
/*Gz
@@ -234,32 +234,32 @@ namespace QCD {
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
fspin(0)=hspin(0);
fspin(1)=hspin(1);
fspin(2)= timesI(hspin(0));
fspin(3)=timesMinusI(hspin(1));
fspin(2)=timesMinusI(hspin(0));
fspin(3)=timesI(hspin(1));
}
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spReconZm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
{
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
fspin(0)=hspin(0);
fspin(1)=hspin(1);
fspin(2)=timesMinusI(hspin(0));
fspin(3)=timesI(hspin(1));
fspin(2)= timesI(hspin(0));
fspin(3)=timesMinusI(hspin(1));
}
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumReconZp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
{
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
fspin(0)+=hspin(0);
fspin(1)+=hspin(1);
fspin(2)+=timesI(hspin(0));
fspin(3)-=timesI(hspin(1));
fspin(2)-=timesI(hspin(0));
fspin(3)+=timesI(hspin(1));
}
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumReconZm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
{
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
fspin(0)+=hspin(0);
fspin(1)+=hspin(1);
fspin(2)-=timesI(hspin(0));
fspin(3)+=timesI(hspin(1));
fspin(2)+=timesI(hspin(0));
fspin(3)-=timesI(hspin(1));
}
/*Gt
* 0 0 1 0 [0]+-[2]
@@ -272,32 +272,32 @@ namespace QCD {
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
fspin(0)=hspin(0);
fspin(1)=hspin(1);
fspin(2)=-hspin(0);
fspin(3)=-hspin(1);
fspin(2)=hspin(0);
fspin(3)=hspin(1);
}
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spReconTm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
{
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
fspin(0)=hspin(0);
fspin(1)=hspin(1);
fspin(2)=hspin(0);
fspin(3)=hspin(1);
fspin(2)=-hspin(0);
fspin(3)=-hspin(1);
}
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumReconTp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
{
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
fspin(0)+=hspin(0);
fspin(1)+=hspin(1);
fspin(2)-=hspin(0);
fspin(3)-=hspin(1);
fspin(2)+=hspin(0);
fspin(3)+=hspin(1);
}
template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumReconTm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
{
//typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
fspin(0)+=hspin(0);
fspin(1)+=hspin(1);
fspin(2)+=hspin(0);
fspin(3)+=hspin(1);
fspin(2)-=hspin(0);
fspin(3)-=hspin(1);
}
/*G5
* 1 0 0 0