ASM instantiation without link errors

Grid/qcd/action/fermion/implementation/StaggeredKernelsAsm.h

@@ -958,12 +958,26 @@ template <> void StaggeredKernels<StaggeredImplD>::DhopSiteAsm(StencilImpl &st,
 #endif
 }
 
-#define KERNEL_INSTANTIATE(CLASS,FUNC,IMPL)			    \
+extern template void StaggeredKernels<StaggeredVec5dImplF>::DhopSiteAsm(StencilImpl &st, LebesgueOrder &lo, 
-  template void CLASS<IMPL>::FUNC(StencilImpl &st, LebesgueOrder &lo,	\
+									DoubledGaugeFieldView &U,
-				  DoubledGaugeFieldView &U,			\
+									DoubledGaugeFieldView &UUU,
-				  DoubledGaugeFieldView &UUU,		\
+									SiteSpinor *buf, int LLs,
-				  SiteSpinor *buf, int LLs,		\
+									int sU, const FermionFieldView &in, FermionFieldView &out,int dag);
-				  int sU, const FermionFieldView &in, FermionFieldView &out,int dag);
+extern template void StaggeredKernels<StaggeredVec5dImplD>::DhopSiteAsm(StencilImpl &st, LebesgueOrder &lo, 
+									DoubledGaugeFieldView &U,
+									DoubledGaugeFieldView &UUU,
+									SiteSpinor *buf, int LLs,
+									int sU, const FermionFieldView &in, FermionFieldView &out,int dag);
+extern template void StaggeredKernels<StaggeredImplF>::DhopSiteAsm(StencilImpl &st, LebesgueOrder &lo, 
+								   DoubledGaugeFieldView &U,
+								   DoubledGaugeFieldView &UUU,
+								   SiteSpinor *buf, int LLs,
+								   int sU, const FermionFieldView &in, FermionFieldView &out,int dag);
+extern template void StaggeredKernels<StaggeredImplD>::DhopSiteAsm(StencilImpl &st, LebesgueOrder &lo, 
+								   DoubledGaugeFieldView &U,
+								   DoubledGaugeFieldView &UUU,
+								   SiteSpinor *buf, int LLs,
+								   int sU, const FermionFieldView &in, FermionFieldView &out,int dag);
 
 NAMESPACE_END(Grid);
 

Grid/qcd/action/fermion/implementation/StaggeredKernelsHand.h

@@ -370,7 +370,6 @@ void StaggeredKernels<Impl>::DhopSiteHandExt(StencilImpl &st, LebesgueOrder &lo,
   }
 }
 
-
 #define DHOP_SITE_HAND_INSTANTIATE(IMPL)				\
   template void StaggeredKernels<IMPL>::DhopSiteHand(StencilImpl &st, LebesgueOrder &lo, \
 						     DoubledGaugeFieldView &U,DoubledGaugeFieldView &UUU, \
@@ -387,6 +386,8 @@ void StaggeredKernels<Impl>::DhopSiteHandExt(StencilImpl &st, LebesgueOrder &lo,
 						     SiteSpinor *buf, int LLs, int sU, \
 						     const FermionFieldView &in, FermionFieldView &out, int dag); \
 
+#undef LOAD_CHI
+
 NAMESPACE_END(Grid);
 
 

Grid/qcd/action/fermion/implementation/WilsonKernelsAsmAvx512.h

@@ -36,7 +36,10 @@ Author: paboyle <paboyle@ph.ed.ac.uk>
     ///////////////////////////////////////////////////////////
 #include <simd/Intel512wilson.h>
 #include <simd/Intel512single.h>
-    
+
+/// Switch off the 5d vectorised code optimisations
+#undef DWFVEC5D
+
 static Vector<vComplexF> signsF;
 
   template<typename vtype>    
@@ -209,6 +212,9 @@ WilsonKernels<ZWilsonImplFH>::AsmDhopSiteDagExt(StencilView &st, DoubledGaugeFie
 /////////////////////////////////////////////////////////////////
 // Ls vectorised, undag Kernel, single
 /////////////////////////////////////////////////////////////////
+
+#ifdef DWFVEC5D
+
 #undef KERNEL_DAG
 #define INTERIOR_AND_EXTERIOR
 #undef INTERIOR
@@ -344,6 +350,8 @@ WilsonKernels<ZDomainWallVec5dImplFH>::AsmDhopSiteDagExt(StencilView &st, Double
 							    int ss,int ssU,int Ls,int Ns,const FermionFieldView &in, FermionFieldView &out)
 #include <qcd/action/fermion/implementation/WilsonKernelsAsmBody.h>
 
+#endif  // VEC 5D
+
 #undef COMPLEX_SIGNS
 #undef MAYBEPERM
 #undef MULT_2SPIN
@@ -510,6 +518,8 @@ WilsonKernels<ZWilsonImplDF>::AsmDhopSiteDagExt(StencilView &st, DoubledGaugeFie
 /////////////////////////////////////////////////////////////////
 // Ls vectorised, undag Kernel, single
 /////////////////////////////////////////////////////////////////
+#ifdef DWFVEC5D
+
 #undef KERNEL_DAG
 #define INTERIOR_AND_EXTERIOR
 #undef INTERIOR
@@ -625,6 +635,7 @@ WilsonKernels<ZDomainWallVec5dImplDF>::AsmDhopSiteDagInt(StencilView &st, Double
 #undef INTERIOR_AND_EXTERIOR
 #undef INTERIOR
 #define EXTERIOR
+
 template<> void 
 WilsonKernels<DomainWallVec5dImplD>::AsmDhopSiteDagExt(StencilView &st, DoubledGaugeFieldView &U,SiteHalfSpinor *buf,
 							    int ss,int ssU,int Ls,int Ns,const FermionFieldView &in, FermionFieldView &out)
@@ -643,8 +654,63 @@ WilsonKernels<ZDomainWallVec5dImplDF>::AsmDhopSiteDagExt(StencilView &st, Double
 							    int ss,int ssU,int Ls,int Ns,const FermionFieldView &in, FermionFieldView &out)
 #include <qcd/action/fermion/implementation/WilsonKernelsAsmBody.h>
 
+#endif  // VEC 5D
+
 #undef COMPLEX_SIGNS
 #undef MAYBEPERM
 #undef MULT_2SPIN
 
+#undef Chi_00
+#undef Chi_01
+#undef Chi_02
+#undef Chi_10
+#undef Chi_11
+#undef Chi_12
+#undef Chi_20
+#undef Chi_21
+#undef Chi_22
+#undef Chi_30
+#undef Chi_31
+#undef Chi_32
+
+#undef UChi_00
+#undef UChi_01
+#undef UChi_02
+#undef UChi_10
+#undef UChi_11
+#undef UChi_12
+#undef UChi_20
+#undef UChi_21
+#undef UChi_22
+#undef UChi_30
+#undef UChi_31
+#undef UChi_32
+
+#undef Psi_00
+#undef Psi_01
+#undef Psi_02
+#undef Psi_10
+#undef Psi_11
+#undef Psi_12
+#undef Psi_20
+#undef Psi_21
+#undef Psi_22
+#undef Psi_30
+#undef Psi_31
+#undef Psi_32
+
+#undef Phi_00
+#undef Phi_01
+#undef Phi_02
+#undef Phi_10
+#undef Phi_11
+#undef Phi_12
+#undef Phi_20
+#undef Phi_21
+#undef Phi_22
+#undef Phi_30
+#undef Phi_31
+#undef Phi_32
+
+
 #endif //AVX512

Grid/qcd/action/fermion/implementation/WilsonKernelsAsmImplementation.h

@@ -38,52 +38,49 @@ NAMESPACE_BEGIN(Grid);
 
 ///////////////////////////////////////////////////////////
 // Default to no assembler implementation
+// Will specialise to 
 ///////////////////////////////////////////////////////////
 template<class Impl> void 
 WilsonKernels<Impl >::AsmDhopSite(StencilView &st, DoubledGaugeFieldView &U,SiteHalfSpinor *buf,
-					  int ss,int ssU,int Ls,int Ns,const FermionFieldView &in, FermionFieldView &out)
+				  int ss,int ssU,int Ls,int Ns,const FermionFieldView &in, FermionFieldView &out)
 {
   assert(0);
 }
 
 template<class Impl> void 
 WilsonKernels<Impl >::AsmDhopSiteDag(StencilView &st, DoubledGaugeFieldView &U,SiteHalfSpinor *buf,
-					     int ss,int ssU,int Ls,int Ns,const FermionFieldView &in, FermionFieldView &out)
+				     int ss,int ssU,int Ls,int Ns,const FermionFieldView &in, FermionFieldView &out)
 {
   assert(0);
 }
 
 template<class Impl> void 
 WilsonKernels<Impl >::AsmDhopSiteInt(StencilView &st, DoubledGaugeFieldView &U,SiteHalfSpinor *buf,
-					  int ss,int ssU,int Ls,int Ns,const FermionFieldView &in, FermionFieldView &out)
+				     int ss,int ssU,int Ls,int Ns,const FermionFieldView &in, FermionFieldView &out)
 {
   assert(0);
 }
 
 template<class Impl> void 
 WilsonKernels<Impl >::AsmDhopSiteDagInt(StencilView &st, DoubledGaugeFieldView &U,SiteHalfSpinor *buf,
-					     int ss,int ssU,int Ls,int Ns,const FermionFieldView &in, FermionFieldView &out)
+					int ss,int ssU,int Ls,int Ns,const FermionFieldView &in, FermionFieldView &out)
 {
   assert(0);
 }
 
 template<class Impl> void 
 WilsonKernels<Impl >::AsmDhopSiteExt(StencilView &st, DoubledGaugeFieldView &U,SiteHalfSpinor *buf,
-					  int ss,int ssU,int Ls,int Ns,const FermionFieldView &in, FermionFieldView &out)
+				     int ss,int ssU,int Ls,int Ns,const FermionFieldView &in, FermionFieldView &out)
 {
   assert(0);
 }
 
 template<class Impl> void 
 WilsonKernels<Impl >::AsmDhopSiteDagExt(StencilView &st, DoubledGaugeFieldView &U,SiteHalfSpinor *buf,
-					     int ss,int ssU,int Ls,int Ns,const FermionFieldView &in, FermionFieldView &out)
+					int ss,int ssU,int Ls,int Ns,const FermionFieldView &in, FermionFieldView &out)
 {
   assert(0);
 }
 
-#include <qcd/action/fermion/implementation/WilsonKernelsAsmAvx512.h>
-#include <qcd/action/fermion/implementation/WilsonKernelsAsmQPX.h>
-
-
 NAMESPACE_END(Grid);
 

Grid/qcd/action/fermion/implementation/WilsonKernelsHandImplementation.h

@@ -31,6 +31,49 @@ Author: paboyle <paboyle@ph.ed.ac.uk>
 
 #include <Grid/qcd/action/fermion/FermionCore.h>
 
+
+#undef LOAD_CHIMU  
+#undef LOAD_CHI 
+#undef MULT_2SPIN
+#undef PERMUTE_DIR
+#undef XP_PROJ  
+#undef YP_PROJ  
+#undef ZP_PROJ  
+#undef TP_PROJ  
+#undef XM_PROJ  
+#undef YM_PROJ  
+#undef ZM_PROJ  
+#undef TM_PROJ  
+#undef XP_RECON 
+#undef XP_RECON_ACCUM 
+#undef XM_RECON 
+#undef XM_RECON_ACCUM 
+#undef YP_RECON_ACCUM 
+#undef YM_RECON_ACCUM 
+#undef ZP_RECON_ACCUM 
+#undef ZM_RECON_ACCUM 
+#undef TP_RECON_ACCUM 
+#undef TM_RECON_ACCUM 
+#undef ZERO_RESULT				 
+#undef Chimu_00
+#undef Chimu_01
+#undef Chimu_02
+#undef Chimu_10
+#undef Chimu_11
+#undef Chimu_12
+#undef Chimu_20
+#undef Chimu_21
+#undef Chimu_22
+#undef Chimu_30
+#undef Chimu_31
+#undef Chimu_32
+#undef HAND_STENCIL_LEG
+#undef HAND_STENCIL_LEG_INT
+#undef HAND_STENCIL_LEG_EXT
+#undef HAND_RESULT
+#undef HAND_RESULT_INT
+#undef HAND_RESULT_EXT
+
 #define REGISTER
 
 #define LOAD_CHIMU \

Grid/qcd/action/fermion/instantiation/StaggeredKernelsInstantiation.cc.master

@@ -29,7 +29,7 @@ directory
 #include <Grid/qcd/action/fermion/FermionCore.h>
 #include <Grid/qcd/action/fermion/implementation/StaggeredKernelsImplementation.h>
 #include <Grid/qcd/action/fermion/implementation/StaggeredKernelsHand.h>
-
+#include <Grid/qcd/action/fermion/implementation/StaggeredKernelsAsm.h>
 
 NAMESPACE_BEGIN(Grid);
 

Grid/qcd/action/fermion/instantiation/WilsonKernelsInstantiation.cc.master

@@ -30,16 +30,11 @@ directory
 /*  END LEGAL */
 #include <Grid/qcd/action/fermion/FermionCore.h>
 #include <Grid/qcd/action/fermion/implementation/WilsonKernelsImplementation.h>
-#include <Grid/qcd/action/fermion/implementation/WilsonKernelsAsmImplementation.h>
 #include <Grid/qcd/action/fermion/implementation/WilsonKernelsGpuImplementation.h>
 #include <Grid/qcd/action/fermion/implementation/WilsonKernelsHandImplementation.h>
-// Do the specialisation only once
-//#include <Grid/qcd/action/fermion/implementation/WilsonKernelsHandGparityImplementation.h>
 
 NAMESPACE_BEGIN(Grid);
 
-// Move these
-
 #include "impl.h"
 template class WilsonKernels<IMPLEMENTATION>; 
 

Grid/qcd/action/fermion/instantiation/WilsonKernelsInstantiationAsm.cc

@@ -0,0 +1,43 @@
+/*************************************************************************************
+
+Grid physics library, www.github.com/paboyle/Grid
+
+Source file: ./lib/qcd/action/fermion/WilsonKernels.cc
+
+Copyright (C) 2015
+
+Author: Peter Boyle <paboyle@ph.ed.ac.uk>
+Author: Peter Boyle <peterboyle@Peters-MacBook-Pro-2.local>
+Author: paboyle <paboyle@ph.ed.ac.uk>
+
+This program is free software; you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation; either version 2 of the License, or
+(at your option) any later version.
+
+This program is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License along
+with this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+
+See the full license in the file "LICENSE" in the top level distribution
+directory
+*************************************************************************************/
+/*  END LEGAL */
+#include <Grid/qcd/action/fermion/FermionCore.h>
+#include <Grid/qcd/action/fermion/implementation/WilsonKernelsImplementation.h>
+#include <Grid/qcd/action/fermion/implementation/WilsonKernelsAsmImplementation.h>
+
+////////////////////////////////////////////////////////////////////////
+// Include the specialisations for ASM kernels
+////////////////////////////////////////////////////////////////////////
+NAMESPACE_BEGIN(Grid);
+#include <Grid/qcd/action/fermion/implementation/WilsonKernelsAsmAvx512.h>
+#include <Grid/qcd/action/fermion/implementation/WilsonKernelsAsmQPX.h>
+NAMESPACE_END(Grid);
+
+