From 72acb0e48f1108a8f8f73289164ffe62bbe75e6a Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@ph.ed.ac.uk>
Date: Sun, 14 Jan 2018 23:41:59 +0000
Subject: [PATCH] Namespace, indent

---
 lib/qcd/action/fermion/CayleyFermion5Dvec.cc | 820 +++++++++----------
 1 file changed, 409 insertions(+), 411 deletions(-)

diff --git a/lib/qcd/action/fermion/CayleyFermion5Dvec.cc b/lib/qcd/action/fermion/CayleyFermion5Dvec.cc
index 653e6ab3..21800523 100644
--- a/lib/qcd/action/fermion/CayleyFermion5Dvec.cc
+++ b/lib/qcd/action/fermion/CayleyFermion5Dvec.cc
@@ -1,4 +1,4 @@
-    /*************************************************************************************
+/*************************************************************************************
 
     Grid physics library, www.github.com/paboyle/Grid 
 
@@ -26,19 +26,19 @@ Author: paboyle <paboyle@ph.ed.ac.uk>
     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 */
+*************************************************************************************/
+/*  END LEGAL */
 
 
 #include <Grid/qcd/action/fermion/FermionCore.h>
 #include <Grid/qcd/action/fermion/CayleyFermion5D.h>
 
 
-namespace Grid {
-namespace QCD {  
-  /*
-   * Dense matrix versions of routines
-   */
+NAMESPACE_BEGIN(Grid);
+
+/*
+ * Dense matrix versions of routines
+ */
 template<class Impl>
 void CayleyFermion5D<Impl>::MooeeInvDag (const FermionField &psi, FermionField &chi)
 {
@@ -79,13 +79,13 @@ void CayleyFermion5D<Impl>::M5D(const FermionField &psi,
   scalar_type * d_p = (scalar_type *)&d[0];
 
   for(int o=0;o<LLs;o++){ // outer
-  for(int i=0;i<nsimd;i++){ //inner
-    int s  = o+i*LLs;
-    int ss = o*nsimd+i;
-    u_p[ss] = upper[s];
-    l_p[ss] = lower[s];
-    d_p[ss] = diag[s];
-  }}
+    for(int i=0;i<nsimd;i++){ //inner
+      int s  = o+i*LLs;
+      int ss = o*nsimd+i;
+      u_p[ss] = upper[s];
+      l_p[ss] = lower[s];
+      d_p[ss] = diag[s];
+    }}
 
 
   M5Dcalls++;
@@ -95,100 +95,100 @@ void CayleyFermion5D<Impl>::M5D(const FermionField &psi,
 
   parallel_for(int ss=0;ss<grid->oSites();ss+=LLs){ // adds LLs
 #if 0
-      alignas(64) SiteHalfSpinor hp;
-      alignas(64) SiteHalfSpinor hm;
-      alignas(64) SiteSpinor fp;
-      alignas(64) SiteSpinor fm;
+    alignas(64) SiteHalfSpinor hp;
+    alignas(64) SiteHalfSpinor hm;
+    alignas(64) SiteSpinor fp;
+    alignas(64) SiteSpinor fm;
 
-      for(int v=0;v<LLs;v++){
+    for(int v=0;v<LLs;v++){
 
-	int vp=(v+1)%LLs;
-	int vm=(v+LLs-1)%LLs;
+      int vp=(v+1)%LLs;
+      int vm=(v+LLs-1)%LLs;
 
-	spProj5m(hp,psi[ss+vp]);
-	spProj5p(hm,psi[ss+vm]);
+      spProj5m(hp,psi[ss+vp]);
+      spProj5p(hm,psi[ss+vm]);
 
-	if ( vp<=v ) rotate(hp,hp,1);
-	if ( vm>=v ) rotate(hm,hm,nsimd-1);
+      if ( vp<=v ) rotate(hp,hp,1);
+      if ( vm>=v ) rotate(hm,hm,nsimd-1);
 	
-	hp=0.5*hp;
-        hm=0.5*hm;
+      hp=0.5*hp;
+      hm=0.5*hm;
 
-	spRecon5m(fp,hp);
-	spRecon5p(fm,hm);
+      spRecon5m(fp,hp);
+      spRecon5p(fm,hm);
 
-	chi[ss+v] = d[v]*phi[ss+v];
-	chi[ss+v] = chi[ss+v]     +u[v]*fp;
-	chi[ss+v] = chi[ss+v]     +l[v]*fm;
+      chi[ss+v] = d[v]*phi[ss+v];
+      chi[ss+v] = chi[ss+v]     +u[v]*fp;
+      chi[ss+v] = chi[ss+v]     +l[v]*fm;
 
-      }
+    }
 #else
-      for(int v=0;v<LLs;v++){
+    for(int v=0;v<LLs;v++){
 
-	vprefetch(psi[ss+v+LLs]);
+      vprefetch(psi[ss+v+LLs]);
 
-	int vp= (v==LLs-1) ? 0     : v+1;
-	int vm= (v==0    ) ? LLs-1 : v-1;
+      int vp= (v==LLs-1) ? 0     : v+1;
+      int vm= (v==0    ) ? LLs-1 : v-1;
 	
-	Simd hp_00 = psi[ss+vp]()(2)(0); 
-	Simd hp_01 = psi[ss+vp]()(2)(1); 
-	Simd hp_02 = psi[ss+vp]()(2)(2); 
-	Simd hp_10 = psi[ss+vp]()(3)(0); 
-	Simd hp_11 = psi[ss+vp]()(3)(1); 
-	Simd hp_12 = psi[ss+vp]()(3)(2); 
+      Simd hp_00 = psi[ss+vp]()(2)(0); 
+      Simd hp_01 = psi[ss+vp]()(2)(1); 
+      Simd hp_02 = psi[ss+vp]()(2)(2); 
+      Simd hp_10 = psi[ss+vp]()(3)(0); 
+      Simd hp_11 = psi[ss+vp]()(3)(1); 
+      Simd hp_12 = psi[ss+vp]()(3)(2); 
 	
-	Simd hm_00 = psi[ss+vm]()(0)(0); 
-	Simd hm_01 = psi[ss+vm]()(0)(1); 
-	Simd hm_02 = psi[ss+vm]()(0)(2); 
-	Simd hm_10 = psi[ss+vm]()(1)(0); 
-	Simd hm_11 = psi[ss+vm]()(1)(1); 
-	Simd hm_12 = psi[ss+vm]()(1)(2); 
-
-	if ( vp<=v ) {
-	  hp_00.v = Optimization::Rotate::tRotate<2>(hp_00.v);
-	  hp_01.v = Optimization::Rotate::tRotate<2>(hp_01.v);
-	  hp_02.v = Optimization::Rotate::tRotate<2>(hp_02.v);
-	  hp_10.v = Optimization::Rotate::tRotate<2>(hp_10.v);
-	  hp_11.v = Optimization::Rotate::tRotate<2>(hp_11.v);
-	  hp_12.v = Optimization::Rotate::tRotate<2>(hp_12.v);
-	}
-	if ( vm>=v ) {
-	  hm_00.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_00.v);
-	  hm_01.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_01.v);
-	  hm_02.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_02.v);
-	  hm_10.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_10.v);
-	  hm_11.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_11.v);
-	  hm_12.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_12.v);
-	}
-
-	// Can force these to real arithmetic and save 2x.
-	Simd p_00  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(0)(0))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_00); 
-	Simd p_01  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(0)(1))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_01); 
-	Simd p_02  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(0)(2))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_02); 
-	Simd p_10  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(1)(0))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_10); 
-	Simd p_11  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(1)(1))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_11); 
-	Simd p_12  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(1)(2))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_12); 
-	Simd p_20  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(2)(0))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_00); 
-	Simd p_21  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(2)(1))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_01); 
-	Simd p_22  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(2)(2))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_02);  
-	Simd p_30  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(3)(0))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_10); 
-	Simd p_31  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(3)(1))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_11); 
-	Simd p_32  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(3)(2))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_12); 
-
-	vstream(chi[ss+v]()(0)(0),p_00);
-	vstream(chi[ss+v]()(0)(1),p_01);
-	vstream(chi[ss+v]()(0)(2),p_02);
-	vstream(chi[ss+v]()(1)(0),p_10);
-	vstream(chi[ss+v]()(1)(1),p_11);
-	vstream(chi[ss+v]()(1)(2),p_12);
-	vstream(chi[ss+v]()(2)(0),p_20);
-	vstream(chi[ss+v]()(2)(1),p_21);
-	vstream(chi[ss+v]()(2)(2),p_22);
-	vstream(chi[ss+v]()(3)(0),p_30);
-	vstream(chi[ss+v]()(3)(1),p_31);
-	vstream(chi[ss+v]()(3)(2),p_32);
+      Simd hm_00 = psi[ss+vm]()(0)(0); 
+      Simd hm_01 = psi[ss+vm]()(0)(1); 
+      Simd hm_02 = psi[ss+vm]()(0)(2); 
+      Simd hm_10 = psi[ss+vm]()(1)(0); 
+      Simd hm_11 = psi[ss+vm]()(1)(1); 
+      Simd hm_12 = psi[ss+vm]()(1)(2); 
 
+      if ( vp<=v ) {
+	hp_00.v = Optimization::Rotate::tRotate<2>(hp_00.v);
+	hp_01.v = Optimization::Rotate::tRotate<2>(hp_01.v);
+	hp_02.v = Optimization::Rotate::tRotate<2>(hp_02.v);
+	hp_10.v = Optimization::Rotate::tRotate<2>(hp_10.v);
+	hp_11.v = Optimization::Rotate::tRotate<2>(hp_11.v);
+	hp_12.v = Optimization::Rotate::tRotate<2>(hp_12.v);
       }
+      if ( vm>=v ) {
+	hm_00.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_00.v);
+	hm_01.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_01.v);
+	hm_02.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_02.v);
+	hm_10.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_10.v);
+	hm_11.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_11.v);
+	hm_12.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_12.v);
+      }
+
+      // Can force these to real arithmetic and save 2x.
+      Simd p_00  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(0)(0))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_00); 
+      Simd p_01  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(0)(1))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_01); 
+      Simd p_02  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(0)(2))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_02); 
+      Simd p_10  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(1)(0))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_10); 
+      Simd p_11  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(1)(1))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_11); 
+      Simd p_12  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(1)(2))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_12); 
+      Simd p_20  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(2)(0))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_00); 
+      Simd p_21  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(2)(1))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_01); 
+      Simd p_22  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(2)(2))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_02);  
+      Simd p_30  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(3)(0))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_10); 
+      Simd p_31  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(3)(1))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_11); 
+      Simd p_32  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(3)(2))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_12); 
+
+      vstream(chi[ss+v]()(0)(0),p_00);
+      vstream(chi[ss+v]()(0)(1),p_01);
+      vstream(chi[ss+v]()(0)(2),p_02);
+      vstream(chi[ss+v]()(1)(0),p_10);
+      vstream(chi[ss+v]()(1)(1),p_11);
+      vstream(chi[ss+v]()(1)(2),p_12);
+      vstream(chi[ss+v]()(2)(0),p_20);
+      vstream(chi[ss+v]()(2)(1),p_21);
+      vstream(chi[ss+v]()(2)(2),p_22);
+      vstream(chi[ss+v]()(3)(0),p_30);
+      vstream(chi[ss+v]()(3)(1),p_31);
+      vstream(chi[ss+v]()(3)(2),p_32);
+
+    }
 #endif
   }
   M5Dtime+=usecond();
@@ -223,13 +223,13 @@ void CayleyFermion5D<Impl>::M5Ddag(const FermionField &psi,
   scalar_type * d_p = (scalar_type *)&d[0];
 
   for(int o=0;o<LLs;o++){ // outer
-  for(int i=0;i<nsimd;i++){ //inner
-    int s  = o+i*LLs;
-    int ss = o*nsimd+i;
-    u_p[ss] = upper[s];
-    l_p[ss] = lower[s];
-    d_p[ss] = diag[s];
-  }}
+    for(int i=0;i<nsimd;i++){ //inner
+      int s  = o+i*LLs;
+      int ss = o*nsimd+i;
+      u_p[ss] = upper[s];
+      l_p[ss] = lower[s];
+      d_p[ss] = diag[s];
+    }}
 
   M5Dcalls++;
   M5Dtime-=usecond();
@@ -261,71 +261,71 @@ void CayleyFermion5D<Impl>::M5Ddag(const FermionField &psi,
 
     }
 #else
-      for(int v=0;v<LLs;v++){
+    for(int v=0;v<LLs;v++){
 
-	vprefetch(psi[ss+v+LLs]);
+      vprefetch(psi[ss+v+LLs]);
 
-	int vp= (v==LLs-1) ? 0     : v+1;
-	int vm= (v==0    ) ? LLs-1 : v-1;
+      int vp= (v==LLs-1) ? 0     : v+1;
+      int vm= (v==0    ) ? LLs-1 : v-1;
 	
-	Simd hp_00 = psi[ss+vp]()(0)(0); 
-	Simd hp_01 = psi[ss+vp]()(0)(1); 
-	Simd hp_02 = psi[ss+vp]()(0)(2); 
-	Simd hp_10 = psi[ss+vp]()(1)(0); 
-	Simd hp_11 = psi[ss+vp]()(1)(1); 
-	Simd hp_12 = psi[ss+vp]()(1)(2); 
+      Simd hp_00 = psi[ss+vp]()(0)(0); 
+      Simd hp_01 = psi[ss+vp]()(0)(1); 
+      Simd hp_02 = psi[ss+vp]()(0)(2); 
+      Simd hp_10 = psi[ss+vp]()(1)(0); 
+      Simd hp_11 = psi[ss+vp]()(1)(1); 
+      Simd hp_12 = psi[ss+vp]()(1)(2); 
 	
-	Simd hm_00 = psi[ss+vm]()(2)(0); 
-	Simd hm_01 = psi[ss+vm]()(2)(1); 
-	Simd hm_02 = psi[ss+vm]()(2)(2); 
-	Simd hm_10 = psi[ss+vm]()(3)(0); 
-	Simd hm_11 = psi[ss+vm]()(3)(1); 
-	Simd hm_12 = psi[ss+vm]()(3)(2); 
+      Simd hm_00 = psi[ss+vm]()(2)(0); 
+      Simd hm_01 = psi[ss+vm]()(2)(1); 
+      Simd hm_02 = psi[ss+vm]()(2)(2); 
+      Simd hm_10 = psi[ss+vm]()(3)(0); 
+      Simd hm_11 = psi[ss+vm]()(3)(1); 
+      Simd hm_12 = psi[ss+vm]()(3)(2); 
 
-	if ( vp<=v ) {
-	  hp_00.v = Optimization::Rotate::tRotate<2>(hp_00.v);
-	  hp_01.v = Optimization::Rotate::tRotate<2>(hp_01.v);
-	  hp_02.v = Optimization::Rotate::tRotate<2>(hp_02.v);
-	  hp_10.v = Optimization::Rotate::tRotate<2>(hp_10.v);
-	  hp_11.v = Optimization::Rotate::tRotate<2>(hp_11.v);
-	  hp_12.v = Optimization::Rotate::tRotate<2>(hp_12.v);
-	}
-	if ( vm>=v ) {
-	  hm_00.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_00.v);
-	  hm_01.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_01.v);
-	  hm_02.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_02.v);
-	  hm_10.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_10.v);
-	  hm_11.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_11.v);
-	  hm_12.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_12.v);
-	}
-
-	Simd p_00  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(0)(0))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_00); 
-	Simd p_01  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(0)(1))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_01); 
-	Simd p_02  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(0)(2))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_02); 
-	Simd p_10  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(1)(0))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_10); 
-	Simd p_11  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(1)(1))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_11); 
-	Simd p_12  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(1)(2))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_12); 
-
-	Simd p_20  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(2)(0))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_00); 
-	Simd p_21  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(2)(1))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_01); 
-	Simd p_22  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(2)(2))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_02);  
-	Simd p_30  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(3)(0))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_10); 
-	Simd p_31  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(3)(1))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_11); 
-	Simd p_32  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(3)(2))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_12); 
-
-	vstream(chi[ss+v]()(0)(0),p_00);
-	vstream(chi[ss+v]()(0)(1),p_01);
-	vstream(chi[ss+v]()(0)(2),p_02);
-	vstream(chi[ss+v]()(1)(0),p_10);
-	vstream(chi[ss+v]()(1)(1),p_11);
-	vstream(chi[ss+v]()(1)(2),p_12);
-	vstream(chi[ss+v]()(2)(0),p_20);
-	vstream(chi[ss+v]()(2)(1),p_21);
-	vstream(chi[ss+v]()(2)(2),p_22);
-	vstream(chi[ss+v]()(3)(0),p_30);
-	vstream(chi[ss+v]()(3)(1),p_31);
-	vstream(chi[ss+v]()(3)(2),p_32);
+      if ( vp<=v ) {
+	hp_00.v = Optimization::Rotate::tRotate<2>(hp_00.v);
+	hp_01.v = Optimization::Rotate::tRotate<2>(hp_01.v);
+	hp_02.v = Optimization::Rotate::tRotate<2>(hp_02.v);
+	hp_10.v = Optimization::Rotate::tRotate<2>(hp_10.v);
+	hp_11.v = Optimization::Rotate::tRotate<2>(hp_11.v);
+	hp_12.v = Optimization::Rotate::tRotate<2>(hp_12.v);
       }
+      if ( vm>=v ) {
+	hm_00.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_00.v);
+	hm_01.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_01.v);
+	hm_02.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_02.v);
+	hm_10.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_10.v);
+	hm_11.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_11.v);
+	hm_12.v = Optimization::Rotate::tRotate<2*Simd::Nsimd()-2>(hm_12.v);
+      }
+
+      Simd p_00  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(0)(0))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_00); 
+      Simd p_01  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(0)(1))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_01); 
+      Simd p_02  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(0)(2))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_02); 
+      Simd p_10  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(1)(0))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_10); 
+      Simd p_11  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(1)(1))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_11); 
+      Simd p_12  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(1)(2))  + switcheroo<Coeff_t>::mult(u[v]()()(),hp_12); 
+
+      Simd p_20  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(2)(0))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_00); 
+      Simd p_21  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(2)(1))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_01); 
+      Simd p_22  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(2)(2))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_02);  
+      Simd p_30  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(3)(0))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_10); 
+      Simd p_31  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(3)(1))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_11); 
+      Simd p_32  = switcheroo<Coeff_t>::mult(d[v]()()(), phi[ss+v]()(3)(2))  + switcheroo<Coeff_t>::mult(l[v]()()(),hm_12); 
+
+      vstream(chi[ss+v]()(0)(0),p_00);
+      vstream(chi[ss+v]()(0)(1),p_01);
+      vstream(chi[ss+v]()(0)(2),p_02);
+      vstream(chi[ss+v]()(1)(0),p_10);
+      vstream(chi[ss+v]()(1)(1),p_11);
+      vstream(chi[ss+v]()(1)(2),p_12);
+      vstream(chi[ss+v]()(2)(0),p_20);
+      vstream(chi[ss+v]()(2)(1),p_21);
+      vstream(chi[ss+v]()(2)(2),p_22);
+      vstream(chi[ss+v]()(3)(0),p_30);
+      vstream(chi[ss+v]()(3)(1),p_31);
+      vstream(chi[ss+v]()(3)(2),p_32);
+    }
 #endif
   }
   M5Dtime+=usecond();
@@ -346,54 +346,54 @@ void CayleyFermion5D<Impl>::MooeeInternalAsm(const FermionField &psi, FermionFie
 {
 #ifndef AVX512
   {
-  SiteHalfSpinor BcastP;
-  SiteHalfSpinor BcastM;
-  SiteHalfSpinor SiteChiP;
-  SiteHalfSpinor SiteChiM;
+    SiteHalfSpinor BcastP;
+    SiteHalfSpinor BcastM;
+    SiteHalfSpinor SiteChiP;
+    SiteHalfSpinor SiteChiM;
 
-  // Ls*Ls * 2 * 12 * vol flops
-  for(int s1=0;s1<LLs;s1++){ 
-    for(int s2=0;s2<LLs;s2++){ 
-      for(int  l=0; l<Simd::Nsimd();l++){ // simd lane
+    // Ls*Ls * 2 * 12 * vol flops
+    for(int s1=0;s1<LLs;s1++){ 
+      for(int s2=0;s2<LLs;s2++){ 
+	for(int  l=0; l<Simd::Nsimd();l++){ // simd lane
 
-        int s=s2+l*LLs;
-	int lex=s2+LLs*site;
+	  int s=s2+l*LLs;
+	  int lex=s2+LLs*site;
 	
-	if ( s2==0 && l==0) {
-	  SiteChiP=zero;
-	  SiteChiM=zero;
-	}
+	  if ( s2==0 && l==0) {
+	    SiteChiP=zero;
+	    SiteChiM=zero;
+	  }
 	
-	for(int sp=0;sp<2;sp++){
-        for(int co=0;co<Nc;co++){
-	  vbroadcast(BcastP()(sp  )(co),psi[lex]()(sp)(co),l);
-	}}
-	for(int sp=0;sp<2;sp++){
-        for(int co=0;co<Nc;co++){
-	  vbroadcast(BcastM()(sp  )(co),psi[lex]()(sp+2)(co),l);
-	}}
+	  for(int sp=0;sp<2;sp++){
+	    for(int co=0;co<Nc;co++){
+	      vbroadcast(BcastP()(sp  )(co),psi[lex]()(sp)(co),l);
+	    }}
+	  for(int sp=0;sp<2;sp++){
+	    for(int co=0;co<Nc;co++){
+	      vbroadcast(BcastM()(sp  )(co),psi[lex]()(sp+2)(co),l);
+	    }}
 
-	for(int sp=0;sp<2;sp++){
-        for(int co=0;co<Nc;co++){
-	  SiteChiP()(sp)(co)=real_madd(Matp[LLs*s+s1]()()(),BcastP()(sp)(co),SiteChiP()(sp)(co)); // 1100 us.
-	  SiteChiM()(sp)(co)=real_madd(Matm[LLs*s+s1]()()(),BcastM()(sp)(co),SiteChiM()(sp)(co)); // each found by commenting out
-	}}
+	  for(int sp=0;sp<2;sp++){
+	    for(int co=0;co<Nc;co++){
+	      SiteChiP()(sp)(co)=real_madd(Matp[LLs*s+s1]()()(),BcastP()(sp)(co),SiteChiP()(sp)(co)); // 1100 us.
+	      SiteChiM()(sp)(co)=real_madd(Matm[LLs*s+s1]()()(),BcastM()(sp)(co),SiteChiM()(sp)(co)); // each found by commenting out
+	    }}
 
-    }}
-    {
-      int lex = s1+LLs*site;
-      for(int sp=0;sp<2;sp++){
-      for(int co=0;co<Nc;co++){
-	vstream(chi[lex]()(sp)(co), SiteChiP()(sp)(co));
-	vstream(chi[lex]()(sp+2)(co), SiteChiM()(sp)(co));
-      }}
+	}}
+      {
+	int lex = s1+LLs*site;
+	for(int sp=0;sp<2;sp++){
+	  for(int co=0;co<Nc;co++){
+	    vstream(chi[lex]()(sp)(co), SiteChiP()(sp)(co));
+	    vstream(chi[lex]()(sp+2)(co), SiteChiM()(sp)(co));
+	  }}
+      }
     }
-  }
 
   }
 #else
   {
-  // pointers
+    // pointers
     //  MASK_REGS;
 #define Chi_00 %%zmm1
 #define Chi_01 %%zmm2
@@ -421,68 +421,68 @@ void CayleyFermion5D<Impl>::MooeeInternalAsm(const FermionField &psi, FermionFie
 #define BCAST10  %%zmm23
 #define BCAST11  %%zmm24
 
-  int incr=LLs*LLs*sizeof(iSinglet<Simd>);
-  for(int s1=0;s1<LLs;s1++){ 
-    for(int s2=0;s2<LLs;s2++){ 
-      int lex=s2+LLs*site;
-      uint64_t a0 = (uint64_t)&Matp[LLs*s2+s1]; // should be cacheable
-      uint64_t a1 = (uint64_t)&Matm[LLs*s2+s1];
-      uint64_t a2 = (uint64_t)&psi[lex];
-      for(int  l=0; l<Simd::Nsimd();l++){ // simd lane
-	if ( (s2+l)==0 ) {
-	  asm (
-  	           VPREFETCH1(0,%2)  	     VPREFETCH1(0,%1)
-  	           VPREFETCH1(12,%2)  	     VPREFETCH1(13,%2)
-  	           VPREFETCH1(14,%2)  	     VPREFETCH1(15,%2)         
-		   VBCASTCDUP(0,%2,BCAST0)   
-		   VBCASTCDUP(1,%2,BCAST1)   
-		   VBCASTCDUP(2,%2,BCAST2)   
-		   VBCASTCDUP(3,%2,BCAST3)   
-		   VBCASTCDUP(4,%2,BCAST4)     VMULMEM (0,%0,BCAST0,Chi_00)
-		   VBCASTCDUP(5,%2,BCAST5)     VMULMEM (0,%0,BCAST1,Chi_01)
-		   VBCASTCDUP(6,%2,BCAST6)     VMULMEM (0,%0,BCAST2,Chi_02)
-		   VBCASTCDUP(7,%2,BCAST7)     VMULMEM (0,%0,BCAST3,Chi_10)
-		   VBCASTCDUP(8,%2,BCAST8)     VMULMEM (0,%0,BCAST4,Chi_11)
-		   VBCASTCDUP(9,%2,BCAST9)     VMULMEM (0,%0,BCAST5,Chi_12)
-		   VBCASTCDUP(10,%2,BCAST10)   VMULMEM (0,%1,BCAST6,Chi_20)
-		   VBCASTCDUP(11,%2,BCAST11)   VMULMEM (0,%1,BCAST7,Chi_21)
-		   VMULMEM (0,%1,BCAST8,Chi_22)         
-		   VMULMEM (0,%1,BCAST9,Chi_30)
-		   VMULMEM (0,%1,BCAST10,Chi_31)       
-		   VMULMEM (0,%1,BCAST11,Chi_32)
-		   : : "r" (a0), "r" (a1), "r" (a2)  );
-	} else { 
-	  asm (
-		   VBCASTCDUP(0,%2,BCAST0)   VMADDMEM (0,%0,BCAST0,Chi_00)
-		   VBCASTCDUP(1,%2,BCAST1)   VMADDMEM (0,%0,BCAST1,Chi_01)
-		   VBCASTCDUP(2,%2,BCAST2)   VMADDMEM (0,%0,BCAST2,Chi_02)
-		   VBCASTCDUP(3,%2,BCAST3)   VMADDMEM (0,%0,BCAST3,Chi_10)
-		   VBCASTCDUP(4,%2,BCAST4)   VMADDMEM (0,%0,BCAST4,Chi_11)
-		   VBCASTCDUP(5,%2,BCAST5)   VMADDMEM (0,%0,BCAST5,Chi_12)
-		   VBCASTCDUP(6,%2,BCAST6)   VMADDMEM (0,%1,BCAST6,Chi_20)
-		   VBCASTCDUP(7,%2,BCAST7)   VMADDMEM (0,%1,BCAST7,Chi_21)
-		   VBCASTCDUP(8,%2,BCAST8)   VMADDMEM (0,%1,BCAST8,Chi_22)
-		   VBCASTCDUP(9,%2,BCAST9)   VMADDMEM (0,%1,BCAST9,Chi_30)
-		   VBCASTCDUP(10,%2,BCAST10)  VMADDMEM (0,%1,BCAST10,Chi_31)
-		   VBCASTCDUP(11,%2,BCAST11)  VMADDMEM (0,%1,BCAST11,Chi_32) 
-		   : : "r" (a0), "r" (a1), "r" (a2)  );
-	}
-	a0 = a0+incr;
-	a1 = a1+incr;
-	a2 = a2+sizeof(Simd::scalar_type);
-      }}
-    {
-      int lexa = s1+LLs*site;
-      asm (
-	       VSTORE(0,%0,Chi_00) VSTORE(1 ,%0,Chi_01)  VSTORE(2 ,%0,Chi_02)		
-	       VSTORE(3,%0,Chi_10) VSTORE(4 ,%0,Chi_11)  VSTORE(5 ,%0,Chi_12)		
-	       VSTORE(6,%0,Chi_20) VSTORE(7 ,%0,Chi_21)  VSTORE(8 ,%0,Chi_22)		
-	       VSTORE(9,%0,Chi_30) VSTORE(10,%0,Chi_31)  VSTORE(11,%0,Chi_32)		
-	       : : "r" ((uint64_t)&chi[lexa]) : "memory" );
+    int incr=LLs*LLs*sizeof(iSinglet<Simd>);
+    for(int s1=0;s1<LLs;s1++){ 
+      for(int s2=0;s2<LLs;s2++){ 
+	int lex=s2+LLs*site;
+	uint64_t a0 = (uint64_t)&Matp[LLs*s2+s1]; // should be cacheable
+	uint64_t a1 = (uint64_t)&Matm[LLs*s2+s1];
+	uint64_t a2 = (uint64_t)&psi[lex];
+	for(int  l=0; l<Simd::Nsimd();l++){ // simd lane
+	  if ( (s2+l)==0 ) {
+	    asm (
+		 VPREFETCH1(0,%2)  	     VPREFETCH1(0,%1)
+		 VPREFETCH1(12,%2)  	     VPREFETCH1(13,%2)
+		 VPREFETCH1(14,%2)  	     VPREFETCH1(15,%2)         
+		 VBCASTCDUP(0,%2,BCAST0)   
+		 VBCASTCDUP(1,%2,BCAST1)   
+		 VBCASTCDUP(2,%2,BCAST2)   
+		 VBCASTCDUP(3,%2,BCAST3)   
+		 VBCASTCDUP(4,%2,BCAST4)     VMULMEM (0,%0,BCAST0,Chi_00)
+		 VBCASTCDUP(5,%2,BCAST5)     VMULMEM (0,%0,BCAST1,Chi_01)
+		 VBCASTCDUP(6,%2,BCAST6)     VMULMEM (0,%0,BCAST2,Chi_02)
+		 VBCASTCDUP(7,%2,BCAST7)     VMULMEM (0,%0,BCAST3,Chi_10)
+		 VBCASTCDUP(8,%2,BCAST8)     VMULMEM (0,%0,BCAST4,Chi_11)
+		 VBCASTCDUP(9,%2,BCAST9)     VMULMEM (0,%0,BCAST5,Chi_12)
+		 VBCASTCDUP(10,%2,BCAST10)   VMULMEM (0,%1,BCAST6,Chi_20)
+		 VBCASTCDUP(11,%2,BCAST11)   VMULMEM (0,%1,BCAST7,Chi_21)
+		 VMULMEM (0,%1,BCAST8,Chi_22)         
+		 VMULMEM (0,%1,BCAST9,Chi_30)
+		 VMULMEM (0,%1,BCAST10,Chi_31)       
+		 VMULMEM (0,%1,BCAST11,Chi_32)
+		 : : "r" (a0), "r" (a1), "r" (a2)  );
+	  } else { 
+	    asm (
+		 VBCASTCDUP(0,%2,BCAST0)   VMADDMEM (0,%0,BCAST0,Chi_00)
+		 VBCASTCDUP(1,%2,BCAST1)   VMADDMEM (0,%0,BCAST1,Chi_01)
+		 VBCASTCDUP(2,%2,BCAST2)   VMADDMEM (0,%0,BCAST2,Chi_02)
+		 VBCASTCDUP(3,%2,BCAST3)   VMADDMEM (0,%0,BCAST3,Chi_10)
+		 VBCASTCDUP(4,%2,BCAST4)   VMADDMEM (0,%0,BCAST4,Chi_11)
+		 VBCASTCDUP(5,%2,BCAST5)   VMADDMEM (0,%0,BCAST5,Chi_12)
+		 VBCASTCDUP(6,%2,BCAST6)   VMADDMEM (0,%1,BCAST6,Chi_20)
+		 VBCASTCDUP(7,%2,BCAST7)   VMADDMEM (0,%1,BCAST7,Chi_21)
+		 VBCASTCDUP(8,%2,BCAST8)   VMADDMEM (0,%1,BCAST8,Chi_22)
+		 VBCASTCDUP(9,%2,BCAST9)   VMADDMEM (0,%1,BCAST9,Chi_30)
+		 VBCASTCDUP(10,%2,BCAST10)  VMADDMEM (0,%1,BCAST10,Chi_31)
+		 VBCASTCDUP(11,%2,BCAST11)  VMADDMEM (0,%1,BCAST11,Chi_32) 
+		 : : "r" (a0), "r" (a1), "r" (a2)  );
+	  }
+	  a0 = a0+incr;
+	  a1 = a1+incr;
+	  a2 = a2+sizeof(Simd::scalar_type);
+	}}
+      {
+	int lexa = s1+LLs*site;
+	asm (
+	     VSTORE(0,%0,Chi_00) VSTORE(1 ,%0,Chi_01)  VSTORE(2 ,%0,Chi_02)		
+	     VSTORE(3,%0,Chi_10) VSTORE(4 ,%0,Chi_11)  VSTORE(5 ,%0,Chi_12)		
+	     VSTORE(6,%0,Chi_20) VSTORE(7 ,%0,Chi_21)  VSTORE(8 ,%0,Chi_22)		
+	     VSTORE(9,%0,Chi_30) VSTORE(10,%0,Chi_31)  VSTORE(11,%0,Chi_32)		
+	     : : "r" ((uint64_t)&chi[lexa]) : "memory" );
 
+      }
     }
   }
-  }
 #undef Chi_00
 #undef Chi_01
 #undef Chi_02
@@ -511,63 +511,63 @@ void CayleyFermion5D<Impl>::MooeeInternalAsm(const FermionField &psi, FermionFie
 #endif
 };
 
-  // Z-mobius version
+// Z-mobius version
 template<class Impl>
 void CayleyFermion5D<Impl>::MooeeInternalZAsm(const FermionField &psi, FermionField &chi,
-					     int LLs, int site, Vector<iSinglet<Simd> > &Matp, Vector<iSinglet<Simd> > &Matm)
+					      int LLs, int site, Vector<iSinglet<Simd> > &Matp, Vector<iSinglet<Simd> > &Matm)
 {
 #ifndef AVX512
   {
-  SiteHalfSpinor BcastP;
-  SiteHalfSpinor BcastM;
-  SiteHalfSpinor SiteChiP;
-  SiteHalfSpinor SiteChiM;
+    SiteHalfSpinor BcastP;
+    SiteHalfSpinor BcastM;
+    SiteHalfSpinor SiteChiP;
+    SiteHalfSpinor SiteChiM;
 
-  // Ls*Ls * 2 * 12 * vol flops
-  for(int s1=0;s1<LLs;s1++){ 
-    for(int s2=0;s2<LLs;s2++){ 
-      for(int  l=0; l<Simd::Nsimd();l++){ // simd lane
+    // Ls*Ls * 2 * 12 * vol flops
+    for(int s1=0;s1<LLs;s1++){ 
+      for(int s2=0;s2<LLs;s2++){ 
+	for(int  l=0; l<Simd::Nsimd();l++){ // simd lane
 
-        int s=s2+l*LLs;
-	int lex=s2+LLs*site;
+	  int s=s2+l*LLs;
+	  int lex=s2+LLs*site;
 	
-	if ( s2==0 && l==0) {
-	  SiteChiP=zero;
-	  SiteChiM=zero;
-	}
+	  if ( s2==0 && l==0) {
+	    SiteChiP=zero;
+	    SiteChiM=zero;
+	  }
 	
-	for(int sp=0;sp<2;sp++){
-        for(int co=0;co<Nc;co++){
-	  vbroadcast(BcastP()(sp  )(co),psi[lex]()(sp)(co),l);
-	}}
-	for(int sp=0;sp<2;sp++){
-        for(int co=0;co<Nc;co++){
-	  vbroadcast(BcastM()(sp  )(co),psi[lex]()(sp+2)(co),l);
-	}}
+	  for(int sp=0;sp<2;sp++){
+	    for(int co=0;co<Nc;co++){
+	      vbroadcast(BcastP()(sp  )(co),psi[lex]()(sp)(co),l);
+	    }}
+	  for(int sp=0;sp<2;sp++){
+	    for(int co=0;co<Nc;co++){
+	      vbroadcast(BcastM()(sp  )(co),psi[lex]()(sp+2)(co),l);
+	    }}
 
-	for(int sp=0;sp<2;sp++){
-        for(int co=0;co<Nc;co++){
-	  SiteChiP()(sp)(co)=SiteChiP()(sp)(co)+ Matp[LLs*s+s1]()()()*BcastP()(sp)(co); 
-	  SiteChiM()(sp)(co)=SiteChiM()(sp)(co)+ Matm[LLs*s+s1]()()()*BcastM()(sp)(co); 
-	}}
+	  for(int sp=0;sp<2;sp++){
+	    for(int co=0;co<Nc;co++){
+	      SiteChiP()(sp)(co)=SiteChiP()(sp)(co)+ Matp[LLs*s+s1]()()()*BcastP()(sp)(co); 
+	      SiteChiM()(sp)(co)=SiteChiM()(sp)(co)+ Matm[LLs*s+s1]()()()*BcastM()(sp)(co); 
+	    }}
 
 
-    }}
-    {
-      int lex = s1+LLs*site;
-      for(int sp=0;sp<2;sp++){
-      for(int co=0;co<Nc;co++){
-	vstream(chi[lex]()(sp)(co), SiteChiP()(sp)(co));
-	vstream(chi[lex]()(sp+2)(co), SiteChiM()(sp)(co));
-      }}
+	}}
+      {
+	int lex = s1+LLs*site;
+	for(int sp=0;sp<2;sp++){
+	  for(int co=0;co<Nc;co++){
+	    vstream(chi[lex]()(sp)(co), SiteChiP()(sp)(co));
+	    vstream(chi[lex]()(sp+2)(co), SiteChiM()(sp)(co));
+	  }}
+      }
     }
-  }
 
   }
 #else
   {
-  // pointers
-  //  MASK_REGS;
+    // pointers
+    //  MASK_REGS;
 #define Chi_00 %zmm0
 #define Chi_01 %zmm1
 #define Chi_02 %zmm2
@@ -611,123 +611,123 @@ void CayleyFermion5D<Impl>::MooeeInternalZAsm(const FermionField &psi, FermionFi
 #define Mm  %zmm26
 #define Mms %zmm27
 #define N 8
-  int incr=LLs*LLs*sizeof(iSinglet<Simd>);
-  for(int s1=0;s1<LLs;s1++){ 
-    for(int s2=0;s2<LLs;s2++){ 
-      int lex=s2+LLs*site;
-      uint64_t a0 = (uint64_t)&Matp[LLs*s2+s1]; // should be cacheable
-      uint64_t a1 = (uint64_t)&Matm[LLs*s2+s1];
-      uint64_t a2 = (uint64_t)&psi[lex];
-      for(int  l=0; l<Simd::Nsimd();l++){ // simd lane
-	if ( (s2+l)==0 ) {
-	  LOAD64(%r8,a0);
-	  LOAD64(%r9,a1);
-	  LOAD64(%r10,a2);
+    int incr=LLs*LLs*sizeof(iSinglet<Simd>);
+    for(int s1=0;s1<LLs;s1++){ 
+      for(int s2=0;s2<LLs;s2++){ 
+	int lex=s2+LLs*site;
+	uint64_t a0 = (uint64_t)&Matp[LLs*s2+s1]; // should be cacheable
+	uint64_t a1 = (uint64_t)&Matm[LLs*s2+s1];
+	uint64_t a2 = (uint64_t)&psi[lex];
+	for(int  l=0; l<Simd::Nsimd();l++){ // simd lane
+	  if ( (s2+l)==0 ) {
+	    LOAD64(%r8,a0);
+	    LOAD64(%r9,a1);
+	    LOAD64(%r10,a2);
+	    asm (
+		 VLOAD(0,%r8,Mp)// i r
+		 VLOAD(0,%r9,Mm)
+		 VSHUF(Mp,Mps)  // r i 
+		 VSHUF(Mm,Mms)
+		 VPREFETCH1(12,%r10)  	     VPREFETCH1(13,%r10)
+		 VPREFETCH1(14,%r10)  	     VPREFETCH1(15,%r10)         
+
+		 VMULIDUP(0*N,%r10,Mps,Chi_00)
+		 VMULIDUP(1*N,%r10,Mps,Chi_01)
+		 VMULIDUP(2*N,%r10,Mps,Chi_02)
+		 VMULIDUP(3*N,%r10,Mps,Chi_10)
+		 VMULIDUP(4*N,%r10,Mps,Chi_11)
+		 VMULIDUP(5*N,%r10,Mps,Chi_12)
+
+		 VMULIDUP(6*N ,%r10,Mms,Chi_20)
+		 VMULIDUP(7*N ,%r10,Mms,Chi_21)
+		 VMULIDUP(8*N ,%r10,Mms,Chi_22)
+		 VMULIDUP(9*N ,%r10,Mms,Chi_30)
+		 VMULIDUP(10*N,%r10,Mms,Chi_31)
+		 VMULIDUP(11*N,%r10,Mms,Chi_32)
+
+		 VMADDSUBRDUP(0*N,%r10,Mp,Chi_00)
+		 VMADDSUBRDUP(1*N,%r10,Mp,Chi_01)
+		 VMADDSUBRDUP(2*N,%r10,Mp,Chi_02)
+		 VMADDSUBRDUP(3*N,%r10,Mp,Chi_10)
+		 VMADDSUBRDUP(4*N,%r10,Mp,Chi_11)
+		 VMADDSUBRDUP(5*N,%r10,Mp,Chi_12)
+
+		 VMADDSUBRDUP(6*N ,%r10,Mm,Chi_20)
+		 VMADDSUBRDUP(7*N ,%r10,Mm,Chi_21)
+		 VMADDSUBRDUP(8*N ,%r10,Mm,Chi_22)
+		 VMADDSUBRDUP(9*N ,%r10,Mm,Chi_30)
+		 VMADDSUBRDUP(10*N,%r10,Mm,Chi_31)
+		 VMADDSUBRDUP(11*N,%r10,Mm,Chi_32)
+		 );
+	  } else { 
+	    LOAD64(%r8,a0);
+	    LOAD64(%r9,a1);
+	    LOAD64(%r10,a2);
+	    asm (
+		 VLOAD(0,%r8,Mp)
+		 VSHUF(Mp,Mps)
+
+		 VLOAD(0,%r9,Mm)
+		 VSHUF(Mm,Mms)
+
+		 VMADDSUBIDUP(0*N,%r10,Mps,Chi_00) //  Mri * Pii +- Cir
+		 VMADDSUBIDUP(1*N,%r10,Mps,Chi_01)
+		 VMADDSUBIDUP(2*N,%r10,Mps,Chi_02)
+		 VMADDSUBIDUP(3*N,%r10,Mps,Chi_10)
+		 VMADDSUBIDUP(4*N,%r10,Mps,Chi_11)
+		 VMADDSUBIDUP(5*N,%r10,Mps,Chi_12)
+
+		 VMADDSUBIDUP(6 *N,%r10,Mms,Chi_20)
+		 VMADDSUBIDUP(7 *N,%r10,Mms,Chi_21)
+		 VMADDSUBIDUP(8 *N,%r10,Mms,Chi_22)
+		 VMADDSUBIDUP(9 *N,%r10,Mms,Chi_30)
+		 VMADDSUBIDUP(10*N,%r10,Mms,Chi_31)
+		 VMADDSUBIDUP(11*N,%r10,Mms,Chi_32)
+
+		 VMADDSUBRDUP(0*N,%r10,Mp,Chi_00) //  Cir = Mir * Prr +- ( Mri * Pii +- Cir) 
+		 VMADDSUBRDUP(1*N,%r10,Mp,Chi_01) //  Ci = MiPr + Ci + MrPi ;    Cr = MrPr - ( MiPi - Cr)
+		 VMADDSUBRDUP(2*N,%r10,Mp,Chi_02)
+		 VMADDSUBRDUP(3*N,%r10,Mp,Chi_10)
+		 VMADDSUBRDUP(4*N,%r10,Mp,Chi_11)
+		 VMADDSUBRDUP(5*N,%r10,Mp,Chi_12)
+
+		 VMADDSUBRDUP(6 *N,%r10,Mm,Chi_20)
+		 VMADDSUBRDUP(7 *N,%r10,Mm,Chi_21)
+		 VMADDSUBRDUP(8 *N,%r10,Mm,Chi_22)
+		 VMADDSUBRDUP(9 *N,%r10,Mm,Chi_30)
+		 VMADDSUBRDUP(10*N,%r10,Mm,Chi_31)
+		 VMADDSUBRDUP(11*N,%r10,Mm,Chi_32)
+		 );
+	  }
+	  a0 = a0+incr;
+	  a1 = a1+incr;
+	  a2 = a2+sizeof(Simd::scalar_type);
+	}}
+      {
+	int lexa = s1+LLs*site;
+	/*
+	  SiteSpinor tmp;
 	  asm (
-	       VLOAD(0,%r8,Mp)// i r
-	       VLOAD(0,%r9,Mm)
-	       VSHUF(Mp,Mps)  // r i 
-	       VSHUF(Mm,Mms)
-	       VPREFETCH1(12,%r10)  	     VPREFETCH1(13,%r10)
-	       VPREFETCH1(14,%r10)  	     VPREFETCH1(15,%r10)         
+	  VSTORE(0,%0,pChi_00) VSTORE(1 ,%0,pChi_01)  VSTORE(2 ,%0,pChi_02)		
+	  VSTORE(3,%0,pChi_10) VSTORE(4 ,%0,pChi_11)  VSTORE(5 ,%0,pChi_12)		
+	  VSTORE(6,%0,pChi_20) VSTORE(7 ,%0,pChi_21)  VSTORE(8 ,%0,pChi_22)		
+	  VSTORE(9,%0,pChi_30) VSTORE(10,%0,pChi_31)  VSTORE(11,%0,pChi_32)		
+	  : : "r" ((uint64_t)&tmp) : "memory" );
+	*/
 
-	       VMULIDUP(0*N,%r10,Mps,Chi_00)
-	       VMULIDUP(1*N,%r10,Mps,Chi_01)
-	       VMULIDUP(2*N,%r10,Mps,Chi_02)
-	       VMULIDUP(3*N,%r10,Mps,Chi_10)
-	       VMULIDUP(4*N,%r10,Mps,Chi_11)
-	       VMULIDUP(5*N,%r10,Mps,Chi_12)
+	asm (
+	     VSTORE(0,%0,pChi_00) VSTORE(1 ,%0,pChi_01)  VSTORE(2 ,%0,pChi_02)		
+	     VSTORE(3,%0,pChi_10) VSTORE(4 ,%0,pChi_11)  VSTORE(5 ,%0,pChi_12)		
+	     VSTORE(6,%0,pChi_20) VSTORE(7 ,%0,pChi_21)  VSTORE(8 ,%0,pChi_22)		
+	     VSTORE(9,%0,pChi_30) VSTORE(10,%0,pChi_31)  VSTORE(11,%0,pChi_32)		
+	     : : "r" ((uint64_t)&chi[lexa]) : "memory" );
 
-	       VMULIDUP(6*N ,%r10,Mms,Chi_20)
-	       VMULIDUP(7*N ,%r10,Mms,Chi_21)
-	       VMULIDUP(8*N ,%r10,Mms,Chi_22)
-	       VMULIDUP(9*N ,%r10,Mms,Chi_30)
-	       VMULIDUP(10*N,%r10,Mms,Chi_31)
-	       VMULIDUP(11*N,%r10,Mms,Chi_32)
-
-	       VMADDSUBRDUP(0*N,%r10,Mp,Chi_00)
-	       VMADDSUBRDUP(1*N,%r10,Mp,Chi_01)
-	       VMADDSUBRDUP(2*N,%r10,Mp,Chi_02)
-	       VMADDSUBRDUP(3*N,%r10,Mp,Chi_10)
-	       VMADDSUBRDUP(4*N,%r10,Mp,Chi_11)
-	       VMADDSUBRDUP(5*N,%r10,Mp,Chi_12)
-
-	       VMADDSUBRDUP(6*N ,%r10,Mm,Chi_20)
-	       VMADDSUBRDUP(7*N ,%r10,Mm,Chi_21)
-	       VMADDSUBRDUP(8*N ,%r10,Mm,Chi_22)
-	       VMADDSUBRDUP(9*N ,%r10,Mm,Chi_30)
-	       VMADDSUBRDUP(10*N,%r10,Mm,Chi_31)
-	       VMADDSUBRDUP(11*N,%r10,Mm,Chi_32)
-	       );
-	} else { 
-	  LOAD64(%r8,a0);
-	  LOAD64(%r9,a1);
-	  LOAD64(%r10,a2);
-	  asm (
-	       VLOAD(0,%r8,Mp)
-	       VSHUF(Mp,Mps)
-
-	       VLOAD(0,%r9,Mm)
-	       VSHUF(Mm,Mms)
-
-	       VMADDSUBIDUP(0*N,%r10,Mps,Chi_00) //  Mri * Pii +- Cir
-	       VMADDSUBIDUP(1*N,%r10,Mps,Chi_01)
-	       VMADDSUBIDUP(2*N,%r10,Mps,Chi_02)
-	       VMADDSUBIDUP(3*N,%r10,Mps,Chi_10)
-	       VMADDSUBIDUP(4*N,%r10,Mps,Chi_11)
-	       VMADDSUBIDUP(5*N,%r10,Mps,Chi_12)
-
-	       VMADDSUBIDUP(6 *N,%r10,Mms,Chi_20)
-	       VMADDSUBIDUP(7 *N,%r10,Mms,Chi_21)
-	       VMADDSUBIDUP(8 *N,%r10,Mms,Chi_22)
-	       VMADDSUBIDUP(9 *N,%r10,Mms,Chi_30)
-	       VMADDSUBIDUP(10*N,%r10,Mms,Chi_31)
-	       VMADDSUBIDUP(11*N,%r10,Mms,Chi_32)
-
-	       VMADDSUBRDUP(0*N,%r10,Mp,Chi_00) //  Cir = Mir * Prr +- ( Mri * Pii +- Cir) 
-	       VMADDSUBRDUP(1*N,%r10,Mp,Chi_01) //  Ci = MiPr + Ci + MrPi ;    Cr = MrPr - ( MiPi - Cr)
-	       VMADDSUBRDUP(2*N,%r10,Mp,Chi_02)
-	       VMADDSUBRDUP(3*N,%r10,Mp,Chi_10)
-	       VMADDSUBRDUP(4*N,%r10,Mp,Chi_11)
-	       VMADDSUBRDUP(5*N,%r10,Mp,Chi_12)
-
-	       VMADDSUBRDUP(6 *N,%r10,Mm,Chi_20)
-	       VMADDSUBRDUP(7 *N,%r10,Mm,Chi_21)
-	       VMADDSUBRDUP(8 *N,%r10,Mm,Chi_22)
-	       VMADDSUBRDUP(9 *N,%r10,Mm,Chi_30)
-	       VMADDSUBRDUP(10*N,%r10,Mm,Chi_31)
-	       VMADDSUBRDUP(11*N,%r10,Mm,Chi_32)
-	       );
-	}
-	a0 = a0+incr;
-	a1 = a1+incr;
-	a2 = a2+sizeof(Simd::scalar_type);
-      }}
-    {
-      int lexa = s1+LLs*site;
-      /*
-      SiteSpinor tmp;
-      asm (
-	       VSTORE(0,%0,pChi_00) VSTORE(1 ,%0,pChi_01)  VSTORE(2 ,%0,pChi_02)		
-	       VSTORE(3,%0,pChi_10) VSTORE(4 ,%0,pChi_11)  VSTORE(5 ,%0,pChi_12)		
-	       VSTORE(6,%0,pChi_20) VSTORE(7 ,%0,pChi_21)  VSTORE(8 ,%0,pChi_22)		
-	       VSTORE(9,%0,pChi_30) VSTORE(10,%0,pChi_31)  VSTORE(11,%0,pChi_32)		
-	       : : "r" ((uint64_t)&tmp) : "memory" );
-      */
-
-      asm (
-	       VSTORE(0,%0,pChi_00) VSTORE(1 ,%0,pChi_01)  VSTORE(2 ,%0,pChi_02)		
-	       VSTORE(3,%0,pChi_10) VSTORE(4 ,%0,pChi_11)  VSTORE(5 ,%0,pChi_12)		
-	       VSTORE(6,%0,pChi_20) VSTORE(7 ,%0,pChi_21)  VSTORE(8 ,%0,pChi_22)		
-	       VSTORE(9,%0,pChi_30) VSTORE(10,%0,pChi_31)  VSTORE(11,%0,pChi_32)		
-	       : : "r" ((uint64_t)&chi[lexa]) : "memory" );
-
-      //      if ( 1 || (site==0) ) { 
-      //	std::cout<<site << " s1 "<<s1<<"\n\t"<<tmp << "\n't" << chi[lexa] <<"\n\t"<<tmp-chi[lexa]<<std::endl;
-      //      }
+	//      if ( 1 || (site==0) ) { 
+	//	std::cout<<site << " s1 "<<s1<<"\n\t"<<tmp << "\n't" << chi[lexa] <<"\n\t"<<tmp-chi[lexa]<<std::endl;
+	//      }
+      }
     }
   }
-  }
 #undef Chi_00
 #undef Chi_01
 #undef Chi_02
@@ -823,6 +823,4 @@ template void CayleyFermion5D<DomainWallVec5dImplDF>::MooeeInternal(const Fermio
 template void CayleyFermion5D<ZDomainWallVec5dImplFH>::MooeeInternal(const FermionField &psi, FermionField &chi,int dag, int inv);
 template void CayleyFermion5D<ZDomainWallVec5dImplDF>::MooeeInternal(const FermionField &psi, FermionField &chi,int dag, int inv);
 
-
-
-}}
+NAMESPACE_END(Grid);