simd 5th dim with rotation

lib/Make.inc

@@ -1,4 +1,4 @@
 
-HFILES=./Algorithms.h ./AlignedAllocator.h ./Cartesian.h ./Communicator.h ./Cshift.h ./Grid.h ./Init.h ./Lattice.h ./Lexicographic.h ./Log.h ./Old/Tensor_peek.h ./Old/Tensor_poke.h ./PerfCount.h ./Simd.h ./Stencil.h ./Tensors.h ./Threads.h ./Timer.h ./algorithms/CoarsenedMatrix.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.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 ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./communicator/Communicator_base.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./lattice/Lattice_ET.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_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 ./parallelIO/BinaryIO.h ./parallelIO/NerscIO.h ./pugixml/pugixml.h ./qcd/QCD.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/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/fermion/WilsonTMFermion.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/gauge/GaugeImpl.h ./qcd/action/gauge/PlaqPlusRectangleAction.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/HmcRunner.h ./qcd/hmc/NerscCheckpointer.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SUn.h ./qcd/utils/SpaceTimeGrid.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/Intel512avx.h ./simd/Intel512avxAddsub.h ./simd/Intel512common.h ./simd/Intel512double.h ./simd/Intel512imci.h ./simd/Intel512single.h ./simd/Intel512wilson.h ./stencil/Lebesgue.h ./tensors/Tensor_Ta.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_trace.h ./tensors/Tensor_traits.h ./tenLine truncated
+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 ./Lexicographic.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/WilsonFermion5DRedBlack.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/fermion/WilsonTMFermion.h ./qcd/action/gauge/GaugeImpl.h ./qcd/action/gauge/PlaqPlusRectangleAction.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/HmcRunner.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/hmc/NerscCheckpointer.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/Intel512avx.h ./simd/Intel512avxAddsub.h ./simd/Intel512common.h ./simd/Intel512double.h ./simd/Intel512imci.h ./simd/Intel512single.h ./simd/Intel512wilson.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_tLine truncated
 
-CCFILES=./Init.cc ./Log.cc ./PerfCount.cc ./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.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/action/fermion/WilsonTMFermion.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/action/fermion/WilsonTMFermion.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

lib/qcd/utils/SpaceTimeGrid.cc

@@ -41,7 +41,11 @@ GridRedBlackCartesian *SpaceTimeGrid::makeFourDimRedBlackGrid(const GridCartesia
 {
   return new GridRedBlackCartesian(FourDimGrid); 
 }
-
+GridCartesian *SpaceTimeGrid::makeFourDimDWFGrid(const std::vector<int> & latt,const std::vector<int> &mpi)
+{
+  std::vector<int> simd(4,1);
+  return makeFourDimGrid(latt,simd,mpi);
+}
 GridCartesian         *SpaceTimeGrid::makeFiveDimGrid(int Ls,const GridCartesian *FourDimGrid)
 {
   int N4=FourDimGrid->_ndimension;
@@ -58,6 +62,7 @@ GridCartesian         *SpaceTimeGrid::makeFiveDimGrid(int Ls,const GridCartesian
   return new GridCartesian(latt5,simd5,mpi5); 
 }
 
+
 GridRedBlackCartesian *SpaceTimeGrid::makeFiveDimRedBlackGrid(int Ls,const GridCartesian *FourDimGrid)
 {
   int N4=FourDimGrid->_ndimension;
@@ -76,4 +81,42 @@ GridRedBlackCartesian *SpaceTimeGrid::makeFiveDimRedBlackGrid(int Ls,const GridC
   return new GridRedBlackCartesian(latt5,simd5,mpi5,cb5,cbd); 
 }
 
+
+GridCartesian         *SpaceTimeGrid::makeFiveDimDWFGrid(int Ls,const GridCartesian *FourDimGrid)
+{
+  int N4=FourDimGrid->_ndimension;
+  int nsimd = FourDimGrid->Nsimd();
+
+  std::vector<int> latt5(1,Ls);
+  std::vector<int> simd5(1,nsimd);
+  std::vector<int>  mpi5(1,1);
+  
+  for(int d=0;d<N4;d++){
+    latt5.push_back(FourDimGrid->_fdimensions[d]);
+    simd5.push_back(1);
+     mpi5.push_back(FourDimGrid->_processors[d]);
+  }
+  return new GridCartesian(latt5,simd5,mpi5); 
+}
+
+GridRedBlackCartesian *SpaceTimeGrid::makeFiveDimDWFRedBlackGrid(int Ls,const GridCartesian *FourDimGrid)
+{
+  int N4=FourDimGrid->_ndimension;
+  int nsimd = FourDimGrid->Nsimd();
+  int cbd=0;
+  std::vector<int> latt5(1,Ls);
+  std::vector<int> simd5(1,nsimd);
+  std::vector<int>  mpi5(1,1);
+  std::vector<int>   cb5(1,1);
+    
+  for(int d=0;d<N4;d++){
+    latt5.push_back(FourDimGrid->_fdimensions[d]);
+    simd5.push_back(1);
+     mpi5.push_back(FourDimGrid->_processors[d]);
+      cb5.push_back(1);
+    }
+  return new GridRedBlackCartesian(latt5,simd5,mpi5,cb5,cbd); 
+}
+
+
 }}

lib/qcd/utils/SpaceTimeGrid.h

@@ -35,9 +35,14 @@ class SpaceTimeGrid {
 
   static GridCartesian         *makeFourDimGrid(const std::vector<int> & latt,const std::vector<int> &simd,const std::vector<int> &mpi);
   static GridRedBlackCartesian *makeFourDimRedBlackGrid       (const GridCartesian *FourDimGrid);
+
   static GridCartesian         *makeFiveDimGrid        (int Ls,const GridCartesian *FourDimGrid);
   static GridRedBlackCartesian *makeFiveDimRedBlackGrid(int Ls,const GridCartesian *FourDimGrid);
 
+  static GridCartesian         *makeFiveDimDWFGrid        (int Ls,const GridCartesian *FourDimGrid);
+  static GridRedBlackCartesian *makeFiveDimDWFRedBlackGrid(int Ls,const GridCartesian *FourDimGrid);
+  static GridCartesian         *makeFourDimDWFGrid        (const std::vector<int> & latt,const std::vector<int> &mpi);
+
 };
 
 }}

lib/tensors/Tensor_extract_merge.h

@@ -44,7 +44,7 @@ template<class vsimd,class scalar>
 inline void extract(typename std::enable_if<!isGridTensor<vsimd>::value, const vsimd >::type * y, 
 		    std::vector<scalar *> &extracted,int offset){
   // FIXME: bounce off memory is painful
-  static const int Nsimd=vsimd::Nsimd();
+  static const int Nsimd=sizeof(vsimd)/sizeof(scalar);
   int Nextr=extracted.size();
   int s=Nsimd/Nextr;
 
@@ -59,7 +59,9 @@ inline void extract(typename std::enable_if<!isGridTensor<vsimd>::value, const v
 template<class vsimd,class scalar>
 inline void merge(typename std::enable_if<!isGridTensor<vsimd>::value, vsimd >::type * y, 
 		  std::vector<scalar *> &extracted,int offset){
-  static const int Nsimd=vsimd::Nsimd();
+
+  static const int Nsimd=sizeof(vsimd)/sizeof(scalar);
+
   int Nextr=extracted.size();
   int s=Nsimd/Nextr; // can have sparse occupation of simd vector if simd_layout does not fill it
                      // replicate n-fold. Use to allow Integer masks to 
@@ -127,7 +129,7 @@ template<class vobj> inline void extract(const vobj &vec,std::vector<typename vo
   typedef typename vobj::scalar_type scalar_type ;
   typedef typename vobj::vector_type vector_type ;
 
-  static const int Nsimd=vobj::vector_type::Nsimd();
+  static const int Nsimd=sizeof(vector_type)/sizeof(scalar_type);
   static const int words=sizeof(vobj)/sizeof(vector_type);
   int Nextr=extracted.size();
   int s=Nsimd/Nextr;
@@ -174,7 +176,7 @@ void merge(vobj &vec,std::vector<typename vobj::scalar_object> &extracted)
   typedef typename vobj::scalar_type scalar_type ;
   typedef typename vobj::vector_type vector_type ;
   
-  static const int Nsimd=vobj::vector_type::Nsimd();
+  static const int Nsimd=sizeof(vector_type)/sizeof(scalar_type);
   static const int words=sizeof(vobj)/sizeof(vector_type);
 
   int Nextr = extracted.size();
@@ -199,7 +201,7 @@ void merge(vobj &vec,std::vector<typename vobj::scalar_object *> &extracted,int
   typedef typename vobj::scalar_type scalar_type ;
   typedef typename vobj::vector_type vector_type ;
   
-  const int Nsimd=vobj::vector_type::Nsimd();
+  const int Nsimd=sizeof(vector_type)/sizeof(scalar_type);
   const int words=sizeof(vobj)/sizeof(vector_type);
 
   int Nextr=extracted.size();