portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-06-13 04:37:05 +01:00
Code Releases Activity

Fix a regression failure on Mobius; chroma regression added

Browse Source
This commit is contained in:
paboyle
2015-12-10 22:55:00 +00:00
parent 26161addd0
commit 3ce10aa975
18 changed files with 132 additions and 57 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
lib/Config.h.in
View File

@ -137,6 +137,9 @@
/* Define to the one symbol short name of this package. */
#undef PACKAGE_TARNAME
/* Define to the home page for this package. */
#undef PACKAGE_URL
/* Define to the version of this package. */
#undef PACKAGE_VERSION

4
lib/Make.inc
View File

@ -1,4 +1,4 @@
HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/ImplicitlyRestartedLanczos.h ./algorithms/iterative/MatrixUtils.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Config.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./Init.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./Log.h ./Old/Tensor_peek.h ./Old/Tensor_poke.h ./parallelIO/BinaryIO.h ./parallelIO/NerscIO.h ./pugixml/pugixml.h ./qcd/action/ActionBase.h ./qcd/action/ActionParams.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/FermionOperatorImpl.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/pseudofermion/EvenOddSchurDifferentiable.h ./qcd/action/pseudofermion/OneFlavourEvenOddRational.h ./qcd/action/pseudofermion/OneFlavourEvenOddRationalRatio.h ./qcd/action/pseudofermion/OneFlavourRational.h ./qcd/action/pseudofermion/OneFlavourRationalRatio.h ./qcd/action/pseudofermion/TwoFlavour.h ./qcd/action/pseudofermion/TwoFlavourEvenOdd.h ./qcd/action/pseudofermion/TwoFlavourEvenOddRatio.h ./qcd/action/pseudofermion/TwoFlavourRatio.h ./qcd/hmc/HMC.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./serialisation/BaseIO.h ./serialisation/BinaryIO.h ./serialisation/MacroMagic.h ./serialisation/Serialisation.h ./serialisation/TextIO.h ./serialisation/XmlIO.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_imci.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_index.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h ./Timer.h
HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/DenseMatrix.h ./algorithms/iterative/EigenSort.h ./algorithms/iterative/Francis.h ./algorithms/iterative/Householder.h ./algorithms/iterative/ImplicitlyRestartedLanczos.h ./algorithms/iterative/Matrix.h ./algorithms/iterative/MatrixUtils.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./Init.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./Log.h ./Old/Tensor_peek.h ./Old/Tensor_poke.h ./parallelIO/BinaryIO.h ./parallelIO/NerscIO.h ./PerfCount.h ./pugixml/pugixml.h ./qcd/action/ActionBase.h ./qcd/action/ActionParams.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/FermionOperatorImpl.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/pseudofermion/EvenOddSchurDifferentiable.h ./qcd/action/pseudofermion/OneFlavourEvenOddRational.h ./qcd/action/pseudofermion/OneFlavourEvenOddRationalRatio.h ./qcd/action/pseudofermion/OneFlavourRational.h ./qcd/action/pseudofermion/OneFlavourRationalRatio.h ./qcd/action/pseudofermion/TwoFlavour.h ./qcd/action/pseudofermion/TwoFlavourEvenOdd.h ./qcd/action/pseudofermion/TwoFlavourEvenOddRatio.h ./qcd/action/pseudofermion/TwoFlavourRatio.h ./qcd/hmc/HMC.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./serialisation/BaseIO.h ./serialisation/BinaryIO.h ./serialisation/MacroMagic.h ./serialisation/Serialisation.h ./serialisation/TextIO.h ./serialisation/XmlIO.h ./simd/Avx512Asm.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_imci.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_index.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h ./Timer.h
CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./Init.cc ./Log.cc ./pugixml/pugixml.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/hmc/HMC.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./serialisation/BinaryIO.cc ./serialisation/TextIO.cc ./serialisation/XmlIO.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./Init.cc ./Log.cc ./PerfCount.cc ./pugixml/pugixml.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsAsm.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/hmc/HMC.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./serialisation/BinaryIO.cc ./serialisation/TextIO.cc ./serialisation/XmlIO.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc

4
lib/PerfCount.cc
View File

@ -3,8 +3,11 @@
#include <PerfCount.h>
namespace Grid {
#define CacheControl(L,O,R) ((PERF_COUNT_HW_CACHE_##L)|(PERF_COUNT_HW_CACHE_OP_##O<<8)| (PERF_COUNT_HW_CACHE_RESULT_##R<<16))
const PerformanceCounter::PerformanceCounterConfig PerformanceCounter::PerformanceCounterConfigs [] = {
#ifdef __linux__
{ PERF_TYPE_HARDWARE, PERF_COUNT_HW_CPU_CYCLES , "CPUCYCLES.........." },
{ PERF_TYPE_HARDWARE, PERF_COUNT_HW_INSTRUCTIONS , "INSTRUCTIONS......." },
{ PERF_TYPE_HARDWARE, PERF_COUNT_HW_CACHE_REFERENCES , "CACHE_REFERENCES..." },
@ -23,6 +26,7 @@ const PerformanceCounter::PerformanceCounterConfig PerformanceCounter::Performan
{ PERF_TYPE_HW_CACHE, CacheControl(LL,PREFETCH,ACCESS) , "LL_PREFETCH_ACCESS."},
{ PERF_TYPE_HW_CACHE, CacheControl(L1I,READ,MISS) , "L1I_READ_MISS......"},
{ PERF_TYPE_HW_CACHE, CacheControl(L1I,READ,ACCESS) , "L1I_READ_ACCESS...."}
#endif
// { PERF_TYPE_HARDWARE, PERF_COUNT_HW_STALLED_CYCLES_FRONTEND, "STALL_CYCLES" },
};
}

1
lib/PerfCount.h
View File

@ -30,6 +30,7 @@ static long perf_event_open(struct perf_event_attr *hw_event, pid_t pid,
class PerformanceCounter {
private:
typedef struct {
public:
uint32_t type;

3
lib/Simd.h
View File

@ -11,7 +11,6 @@
// Vector types are arch dependent
////////////////////////////////////////////////////////////////////////
typedef uint32_t Integer;
#define _MM_SELECT_FOUR_FOUR(A,B,C,D) ((A<<6)|(B<<4)|(C<<2)|(D))
#define _MM_SELECT_EIGHT_TWO(A,B,C,D,E,F,G,H) ((A<<7)|(B<<6)|(C<<5)|(D<<4)|(E<<3)|(F<<2)|(G<<4)|(H))
@ -20,6 +19,8 @@ typedef uint32_t Integer;
namespace Grid {
typedef uint32_t Integer;
typedef float RealF;
typedef double RealD;
#ifdef GRID_DEFAULT_PRECISION_DOUBLE

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);
}}

31
lib/qcd/action/fermion/WilsonKernels.h
View File

@ -28,19 +28,12 @@ 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
void 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
#define HANDOPT
#ifdef HANDOPT
void DiracOptHandDhopSite(StencilImpl &st,DoubledGaugeField &U,
std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> > &buf,
int sF,int sU,const FermionField &in, FermionField &out);
@ -48,25 +41,9 @@ 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);
}}

5
lib/simd/Grid_avx.h
View File

@ -16,6 +16,7 @@
#define _mm256_set_m128i(hi,lo) _mm256_insertf128_si256(_mm256_castsi128_si256(lo),(hi),1)
#endif
namespace Grid {
namespace Optimization {
template<class vtype>
@ -467,7 +468,7 @@ namespace Optimization {
//////////////////////////////////////////////////////////////////////////////////////
// Here assign types
namespace Grid {
typedef __m256 SIMD_Ftype; // Single precision type
typedef __m256d SIMD_Dtype; // Double precision type
typedef __m256i SIMD_Itype; // Integer type
@ -488,8 +489,8 @@ namespace Grid {
typedef Optimization::Vstore VstoreSIMD;
typedef Optimization::Vset VsetSIMD;
typedef Optimization::Vstream VstreamSIMD;
template <typename S, typename T> using ReduceSIMD = Optimization::Reduce<S,T>;
template <typename S, typename T> using ReduceSIMD = Optimization::Reduce<S,T>;
// Arithmetic operations
typedef Optimization::Sum SumSIMD;