portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-04-06 04:05:55 +01:00
Code Releases Activity

Merge branch 'develop' of github.com:fionnoh/Grid into feature/A2A_current_insertion

Browse Source
This commit is contained in:
Fionn O hOgain 2019-08-07 12:11:40 +01:00
commit d566637cec
17 changed files with 110 additions and 192 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
Grid/qcd/action/fermion/ImprovedStaggeredFermion.cc
View File

@ -26,7 +26,7 @@ See the full license in the file "LICENSE" in the top level distribution
directory
*************************************************************************************/
/* END LEGAL */
#include <Grid.h>
#include <Grid/Grid.h>
namespace Grid {
namespace QCD {

14
Grid/qcd/action/fermion/StaggeredKernelsAsm.cc
View File

@ -26,11 +26,11 @@ Author: paboyle <paboyle@ph.ed.ac.uk>
See the full license in the file "LICENSE" in the top level distribution directory
*************************************************************************************/
/* END LEGAL */
#include <Grid.h>
#include <Grid/Grid.h>
#ifdef AVX512
#include <simd/Intel512common.h>
#include <simd/Intel512avx.h>
#include <Grid/simd/Intel512common.h>
#include <Grid/simd/Intel512avx.h>
#endif
// Interleave operations from two directions
@ -679,7 +679,7 @@ void StaggeredKernels<Impl>::DhopSiteAsm(StencilImpl &st, LebesgueOrder &lo,
gauge3 =(uint64_t)&UU._odata[sU]( T );
// This is the single precision 5th direction vectorised kernel
#include <simd/Intel512single.h>
#include <Grid/simd/Intel512single.h>
template <> void StaggeredKernels<StaggeredVec5dImplF>::DhopSiteAsm(StencilImpl &st, LebesgueOrder &lo,
DoubledGaugeField &U, DoubledGaugeField &UUU,
SiteSpinor *buf, int LLs, int sU,
@ -732,7 +732,7 @@ template <> void StaggeredKernels<StaggeredVec5dImplF>::DhopSiteAsm(StencilImpl
}
#include <simd/Intel512double.h>
#include <Grid/simd/Intel512double.h>
template <> void StaggeredKernels<StaggeredVec5dImplD>::DhopSiteAsm(StencilImpl &st, LebesgueOrder &lo,
DoubledGaugeField &U, DoubledGaugeField &UUU,
SiteSpinor *buf, int LLs, int sU,
@ -816,7 +816,7 @@ template <> void StaggeredKernels<StaggeredVec5dImplD>::DhopSiteAsm(StencilImpl
// This is the single precision 5th direction vectorised kernel
#include <simd/Intel512single.h>
#include <Grid/simd/Intel512single.h>
template <> void StaggeredKernels<StaggeredImplF>::DhopSiteAsm(StencilImpl &st, LebesgueOrder &lo,
DoubledGaugeField &U, DoubledGaugeField &UUU,
SiteSpinor *buf, int LLs, int sU,
@ -884,7 +884,7 @@ template <> void StaggeredKernels<StaggeredImplF>::DhopSiteAsm(StencilImpl &st,
#endif
}
#include <simd/Intel512double.h>
#include <Grid/simd/Intel512double.h>
template <> void StaggeredKernels<StaggeredImplD>::DhopSiteAsm(StencilImpl &st, LebesgueOrder &lo,
DoubledGaugeField &U, DoubledGaugeField &UUU,
SiteSpinor *buf, int LLs, int sU,

2
Grid/qcd/action/fermion/StaggeredKernelsHand.cc
View File

@ -26,7 +26,7 @@ Author: paboyle <paboyle@ph.ed.ac.uk>
See the full license in the file "LICENSE" in the top level distribution directory
*************************************************************************************/
/* END LEGAL */
#include <Grid.h>
#include <Grid/Grid.h>
#define LOAD_CHI(b) \

4
Grid/qcd/action/fermion/WilsonKernelsAsm.cc
View File

@ -81,8 +81,8 @@ WilsonKernels<Impl >::AsmDhopSiteDagExt(StencilImpl &st,LebesgueOrder & lo,Doubl
assert(0);
}
#include <qcd/action/fermion/WilsonKernelsAsmAvx512.h>
#include <qcd/action/fermion/WilsonKernelsAsmQPX.h>
#include <Grid/qcd/action/fermion/WilsonKernelsAsmAvx512.h>
#include <Grid/qcd/action/fermion/WilsonKernelsAsmQPX.h>
#define INSTANTIATE_ASM(A)\
template void WilsonKernels<A>::AsmDhopSite(StencilImpl &st,LebesgueOrder & lo,DoubledGaugeField &U, SiteHalfSpinor *buf,\

1
Grid/qcd/action/pseudofermion/ExactOneFlavourRatio.h
View File

@ -66,6 +66,7 @@ namespace QCD{
FermionField Phi; // the pseudofermion field for this trajectory
public:
ExactOneFlavourRatioPseudoFermionAction(AbstractEOFAFermion<Impl>& _Lop,
AbstractEOFAFermion<Impl>& _Rop,
OperatorFunction<FermionField>& HeatbathCG,

79
Grid/simd/Grid_avx512.h
View File

@ -485,83 +485,6 @@ namespace Optimization {
// Some Template specialization
// Hack for CLANG until mm512_reduce_add_ps etc... are implemented in GCC and Clang releases
#ifndef __INTEL_COMPILER
#warning "Slow reduction due to incomplete reduce intrinsics"
//Complex float Reduce
template<>
inline Grid::ComplexF Reduce<Grid::ComplexF, __m512>::operator()(__m512 in){
__m512 v1,v2;
v1=Optimization::Permute::Permute0(in); // avx 512; quad complex single
v1= _mm512_add_ps(v1,in);
v2=Optimization::Permute::Permute1(v1);
v1 = _mm512_add_ps(v1,v2);
v2=Optimization::Permute::Permute2(v1);
v1 = _mm512_add_ps(v1,v2);
u512f conv; conv.v = v1;
return Grid::ComplexF(conv.f[0],conv.f[1]);
}
//Real float Reduce
template<>
inline Grid::RealF Reduce<Grid::RealF, __m512>::operator()(__m512 in){
__m512 v1,v2;
v1 = Optimization::Permute::Permute0(in); // avx 512; octo-double
v1 = _mm512_add_ps(v1,in);
v2 = Optimization::Permute::Permute1(v1);
v1 = _mm512_add_ps(v1,v2);
v2 = Optimization::Permute::Permute2(v1);
v1 = _mm512_add_ps(v1,v2);
v2 = Optimization::Permute::Permute3(v1);
v1 = _mm512_add_ps(v1,v2);
u512f conv; conv.v=v1;
return conv.f[0];
}
//Complex double Reduce
template<>
inline Grid::ComplexD Reduce<Grid::ComplexD, __m512d>::operator()(__m512d in){
__m512d v1;
v1 = Optimization::Permute::Permute0(in); // sse 128; paired complex single
v1 = _mm512_add_pd(v1,in);
v1 = Optimization::Permute::Permute1(in); // sse 128; paired complex single
v1 = _mm512_add_pd(v1,in);
u512d conv; conv.v = v1;
return Grid::ComplexD(conv.f[0],conv.f[1]);
}
//Real double Reduce
template<>
inline Grid::RealD Reduce<Grid::RealD, __m512d>::operator()(__m512d in){
__m512d v1,v2;
v1 = Optimization::Permute::Permute0(in); // avx 512; quad double
v1 = _mm512_add_pd(v1,in);
v2 = Optimization::Permute::Permute1(v1);
v1 = _mm512_add_pd(v1,v2);
v2 = Optimization::Permute::Permute2(v1);
v1 = _mm512_add_pd(v1,v2);
u512d conv; conv.v = v1;
return conv.f[0];
}
//Integer Reduce
template<>
inline Integer Reduce<Integer, __m512i>::operator()(__m512i in){
// No full vector reduce, use AVX to add upper and lower halves of register
// and perform AVX reduction.
__m256i v1, v2, v3;
__m128i u1, u2, ret;
v1 = _mm512_castsi512_si256(in); // upper half
v2 = _mm512_extracti32x8_epi32(in, 1); // lower half
v3 = _mm256_add_epi32(v1, v2);
v1 = _mm256_hadd_epi32(v3, v3);
v2 = _mm256_hadd_epi32(v1, v1);
u1 = _mm256_castsi256_si128(v2); // upper half
u2 = _mm256_extracti128_si256(v2, 1); // lower half
ret = _mm_add_epi32(u1, u2);
return _mm_cvtsi128_si32(ret);
}
#else
//Complex float Reduce
template<>
inline Grid::ComplexF Reduce<Grid::ComplexF, __m512>::operator()(__m512 in){
@ -590,8 +513,6 @@ namespace Optimization {
inline Integer Reduce<Integer, __m512i>::operator()(__m512i in){
return _mm512_reduce_add_epi32(in);
}
#endif
}

3
HMC/Mobius2p1fEOFA.cc
View File

@ -30,7 +30,9 @@ directory
/* END LEGAL */
#include <Grid/Grid.h>
#ifdef GRID_DEFAULT_PRECISION_DOUBLE
#define MIXED_PRECISION
#endif
namespace Grid{
namespace QCD{
@ -346,6 +348,7 @@ int main(int argc, char **argv) {
#else
ExactOneFlavourRatioPseudoFermionAction<FermionImplPolicy>
EOFA(Strange_Op_L, Strange_Op_R,
ActionCG,
ActionCG, ActionCG,
DerivativeCG, DerivativeCG,
OFRp, true);

9
Hadrons/Global.hpp
View File

@ -109,10 +109,11 @@ typedef std::vector<typename ComplexField##suffix::vector_object::scalar_object>
#define FERM_TYPE_ALIASES(FImpl, suffix)\
BASIC_TYPE_ALIASES(FImpl, suffix);\
typedef FermionOperator<FImpl> FMat##suffix;\
typedef typename FImpl::FermionField FermionField##suffix;\
typedef typename FImpl::GaugeField GaugeField##suffix;\
typedef typename FImpl::DoubledGaugeField DoubledGaugeField##suffix;
typedef FermionOperator<FImpl> FMat##suffix;\
typedef typename FImpl::FermionField FermionField##suffix;\
typedef typename FImpl::GaugeField GaugeField##suffix;\
typedef typename FImpl::DoubledGaugeField DoubledGaugeField##suffix;\
typedef Lattice<iSpinMatrix<typename FImpl::Simd>> SpinMatrixField##suffix;
#define GAUGE_TYPE_ALIASES(GImpl, suffix)\
typedef typename GImpl::GaugeField GaugeField##suffix;

79
Hadrons/Modules/MContraction/WeakMesonDecayKl2.hpp
View File

@ -64,7 +64,7 @@ BEGIN_HADRONS_NAMESPACE
*/
/******************************************************************************
* TWeakMesonDecayKl2 *
* TWeakMesonDecayKl2 *
******************************************************************************/
BEGIN_MODULE_NAMESPACE(MContraction)
@ -75,7 +75,7 @@ public:
std::string, q1,
std::string, q2,
std::string, lepton,
std::string, output);
std::string, output);
};
template <typename FImpl>
@ -83,14 +83,13 @@ class TWeakMesonDecayKl2: public Module<WeakMesonDecayKl2Par>
{
public:
FERM_TYPE_ALIASES(FImpl,);
class Metadata: Serializable
typedef typename SpinMatrixField::vector_object::scalar_object SpinMatrix;
class Result: Serializable
{
public:
GRID_SERIALIZABLE_CLASS_MEMBERS(Metadata,
int, spinidx1,
int, spinidx2);
GRID_SERIALIZABLE_CLASS_MEMBERS(Result,
std::vector<SpinMatrix>, corr);
};
typedef Correlator<Metadata> Result;
public:
// constructor
TWeakMesonDecayKl2(const std::string name);
@ -138,10 +137,10 @@ std::vector<std::string> TWeakMesonDecayKl2<FImpl>::getOutput(void)
template <typename FImpl>
void TWeakMesonDecayKl2<FImpl>::setup(void)
{
envTmpLat(LatticeComplex, "c");
envTmpLat(ComplexField, "c");
envTmpLat(PropagatorField, "prop_buf");
envCreateLat(PropagatorField, getName());
envTmpLat(LatticeComplex, "buf");
envTmpLat(SpinMatrixField, "buf");
}
// execution ///////////////////////////////////////////////////////////////////
@ -150,57 +149,33 @@ void TWeakMesonDecayKl2<FImpl>::execute(void)
{
LOG(Message) << "Computing QED Kl2 contractions '" << getName() << "' using"
<< " quarks '" << par().q1 << "' and '" << par().q2 << "' and"
<< "lepton '" << par().lepton << "'" << std::endl;
<< "lepton '" << par().lepton << "'" << std::endl;
Gamma g5(Gamma::Algebra::Gamma5);
int nt = env().getDim(Tp);
std::vector<SpinMatrix> res_summed;
Result r;
auto &res = envGet(PropagatorField, getName()); res = zero;
Gamma g5(Gamma::Algebra::Gamma5);
int nt = env().getDim(Tp);
auto &q1 = envGet(PropagatorField, par().q1);
auto &q2 = envGet(PropagatorField, par().q2);
auto &res = envGet(PropagatorField, getName()); res = zero;
auto &q1 = envGet(PropagatorField, par().q1);
auto &q2 = envGet(PropagatorField, par().q2);
auto &lepton = envGet(PropagatorField, par().lepton);
envGetTmp(LatticeComplex, buf);
std::vector<TComplex> res_summed;
envGetTmp(LatticeComplex, c);
envGetTmp(SpinMatrixField, buf);
envGetTmp(ComplexField, c);
envGetTmp(PropagatorField, prop_buf);
std::vector<Result> result;
Result r;
for (unsigned int mu = 0; mu < 4; ++mu)
{
c = zero;
//hadronic part: trace(q1*adj(q2)*g5*gL[mu])
c = trace(q1*adj(q2)*g5*GammaL(Gamma::gmu[mu]));
prop_buf = 1.;
//multiply lepton part
res += c * prop_buf * GammaL(Gamma::gmu[mu]) * lepton;
c = zero;
//hadronic part: trace(q1*adj(q2)*g5*gL[mu])
c = trace(q1*adj(q2)*g5*GammaL(Gamma::gmu[mu]));
prop_buf = 1.;
//multiply lepton part
res += c * prop_buf * GammaL(Gamma::gmu[mu]) * lepton;
}
//loop over spinor index of lepton part
unsigned int i = 0;
for (unsigned int s1 = 0; s1 < Ns ; ++s1)
for (unsigned int s2 = 0; s2 < Ns ; ++s2)
{
buf = peekColour(peekSpin(res,s1,s2),0,0);
sliceSum(buf, res_summed, Tp);
r.corr.clear();
for (unsigned int t = 0; t < nt; ++t)
{
r.corr.push_back(TensorRemove(res_summed[t]));
}
r.info.spinidx1 = s1;
r.info.spinidx2 = s2;
result.push_back(r);
i+=1;
}
saveResult(par().output, "weakdecay", result);
buf = peekColour(res, 0, 0);
sliceSum(buf, r.corr, Tp);
saveResult(par().output, "weakdecay", r);
}
END_MODULE_NAMESPACE

53
Hadrons/Modules/MFermion/EMLepton.hpp
View File

@ -54,8 +54,9 @@ BEGIN_HADRONS_NAMESPACE
* - action: fermion action used for propagator (string)
* - emField: photon field A_mu (string)
* - mass: input mass for the lepton propagator
* - boundary: boundary conditions for the lepton propagator, e.g. "1 1 1 -1"
* - twist: twisted boundary for lepton propagator, e.g. "0.0 0.0 0.0 0.5"
* - deltat: source-sink separation
* - deltat: list of source-sink separations
*
*******************************************************************************/
@ -74,7 +75,7 @@ public:
double, mass,
std::string , boundary,
std::string, twist,
unsigned int, deltat);
std::vector<unsigned int>, deltat);
};
template <typename FImpl>
@ -124,7 +125,12 @@ std::vector<std::string> TEMLepton<FImpl>::getInput(void)
template <typename FImpl>
std::vector<std::string> TEMLepton<FImpl>::getOutput(void)
{
std::vector<std::string> out = {getName(), getName() + "_free"};
std::vector<std::string> out = {};
for(int i=0; i<par().deltat.size(); i++)
{
out.push_back(std::to_string(par().deltat[i]) + "_" + getName() + "_free");
out.push_back(std::to_string(par().deltat[i]) + "_" + getName());
}
return out;
}
@ -134,8 +140,11 @@ template <typename FImpl>
void TEMLepton<FImpl>::setup(void)
{
Ls_ = env().getObjectLs(par().action);
envCreateLat(PropagatorField, getName());
envCreateLat(PropagatorField, getName() + "_free");
for(int i=0; i<par().deltat.size(); i++)
{
envCreateLat(PropagatorField, std::to_string(par().deltat[i]) + "_" + getName() + "_free");
envCreateLat(PropagatorField, std::to_string(par().deltat[i]) + "_" + getName());
}
envTmpLat(FermionField, "source", Ls_);
envTmpLat(FermionField, "sol", Ls_);
envTmpLat(FermionField, "tmp");
@ -156,9 +165,6 @@ void TEMLepton<FImpl>::execute(void)
auto &mat = envGet(FMat, par().action);
RealD mass = par().mass;
Complex ci(0.0,1.0);
PropagatorField &Aslashlep = envGet(PropagatorField, getName());
PropagatorField &lep = envGet(PropagatorField, getName() + "_free");
envGetTmp(FermionField, source);
envGetTmp(FermionField, sol);
@ -227,6 +233,22 @@ void TEMLepton<FImpl>::execute(void)
}
}
for(unsigned int dt=0;dt<par().deltat.size();dt++){
PropagatorField &lep = envGet(PropagatorField, std::to_string(par().deltat[dt]) + "_" + getName() + "_free");
for(tl=0;tl<nt;tl++){
//shift free propagator to different source positions
//account for possible anti-periodic boundary in time
proptmp = Cshift(freetmp,Tp, -tl);
proptmp = where( tlat < tl, boundary[Tp]*proptmp, proptmp);
// free propagator for fixed source-sink separation
lep = where(tlat == (tl-par().deltat[dt]+nt)%nt, proptmp, lep);
}
//account for possible anti-periodic boundary in time
lep = where( tlat >= nt-par().deltat[dt], boundary[Tp]*lep, lep);
}
for(tl=0;tl<nt;tl++){
//shift free propagator to different source positions
@ -234,9 +256,6 @@ void TEMLepton<FImpl>::execute(void)
proptmp = Cshift(freetmp,Tp, -tl);
proptmp = where( tlat < tl, boundary[Tp]*proptmp, proptmp);
// free propagator for fixed source-sink separation
lep = where(tlat == (tl-par().deltat+nt)%nt, proptmp, lep);
// i*A_mu*gamma_mu
sourcetmp = zero;
for(unsigned int mu=0;mu<=3;mu++)
@ -276,13 +295,17 @@ void TEMLepton<FImpl>::execute(void)
}
}
// keep the result for the desired delta t
Aslashlep = where(tlat == (tl-par().deltat+nt)%nt, proptmp, Aslashlep);
for(unsigned int dt=0;dt<par().deltat.size();dt++){
PropagatorField &Aslashlep = envGet(PropagatorField, std::to_string(par().deltat[dt]) + "_" + getName());
Aslashlep = where(tlat == (tl-par().deltat[dt]+nt)%nt, proptmp, Aslashlep);
}
}
//account for possible anti-periodic boundary in time
Aslashlep = where( tlat >= nt-par().deltat, boundary[Tp]*Aslashlep, Aslashlep);
lep = where( tlat >= nt-par().deltat, boundary[Tp]*lep, lep);
for(unsigned int dt=0;dt<par().deltat.size();dt++){
PropagatorField &Aslashlep = envGet(PropagatorField, std::to_string(par().deltat[dt]) + "_" + getName());
Aslashlep = where( tlat >= nt-par().deltat[dt], boundary[Tp]*Aslashlep, Aslashlep);
}
}
END_MODULE_NAMESPACE

33
benchmarks/Benchmark_comms.cc
View File

@ -188,9 +188,9 @@ int main (int argc, char ** argv)
rbuf[mu].resize(lat*lat*lat*Ls);
// std::cout << " buffers " << std::hex << (uint64_t)&xbuf[mu][0] <<" " << (uint64_t)&rbuf[mu][0] <<std::endl;
}
uint64_t bytes=lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD);
int ncomm;
int bytes=lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD);
for(int i=0;i<Nloop;i++){
double start=usecond();
@ -277,15 +277,15 @@ int main (int argc, char ** argv)
std::vector<HalfSpinColourVectorD *> xbuf(8);
std::vector<HalfSpinColourVectorD *> rbuf(8);
Grid.ShmBufferFreeAll();
uint64_t bytes = lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD);
for(int d=0;d<8;d++){
xbuf[d] = (HalfSpinColourVectorD *)Grid.ShmBufferMalloc(lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD));
rbuf[d] = (HalfSpinColourVectorD *)Grid.ShmBufferMalloc(lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD));
bzero((void *)xbuf[d],lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD));
bzero((void *)rbuf[d],lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD));
xbuf[d] = (HalfSpinColourVectorD *)Grid.ShmBufferMalloc(bytes);
rbuf[d] = (HalfSpinColourVectorD *)Grid.ShmBufferMalloc(bytes);
bzero((void *)xbuf[d],bytes);
bzero((void *)rbuf[d],bytes);
}
int ncomm;
int bytes=lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD);
double dbytes;
for(int i=0;i<Nloop;i++){
@ -374,15 +374,15 @@ int main (int argc, char ** argv)
std::vector<HalfSpinColourVectorD *> xbuf(8);
std::vector<HalfSpinColourVectorD *> rbuf(8);
Grid.ShmBufferFreeAll();
uint64_t bytes=lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD);
for(int d=0;d<8;d++){
xbuf[d] = (HalfSpinColourVectorD *)Grid.ShmBufferMalloc(lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD));
rbuf[d] = (HalfSpinColourVectorD *)Grid.ShmBufferMalloc(lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD));
bzero((void *)xbuf[d],lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD));
bzero((void *)rbuf[d],lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD));
xbuf[d] = (HalfSpinColourVectorD *)Grid.ShmBufferMalloc(bytes);
rbuf[d] = (HalfSpinColourVectorD *)Grid.ShmBufferMalloc(bytes);
bzero((void *)xbuf[d],bytes);
bzero((void *)rbuf[d],bytes);
}
int ncomm;
int bytes=lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD);
double dbytes;
for(int i=0;i<Nloop;i++){
double start=usecond();
@ -472,15 +472,16 @@ int main (int argc, char ** argv)
std::vector<HalfSpinColourVectorD *> xbuf(8);
std::vector<HalfSpinColourVectorD *> rbuf(8);
Grid.ShmBufferFreeAll();
// uint64_t bytes = lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD);
uint64_t bytes = 2*1024*1024;
for(int d=0;d<8;d++){
xbuf[d] = (HalfSpinColourVectorD *)Grid.ShmBufferMalloc(lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD));
rbuf[d] = (HalfSpinColourVectorD *)Grid.ShmBufferMalloc(lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD));
bzero((void *)xbuf[d],lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD));
bzero((void *)rbuf[d],lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD));
xbuf[d] = (HalfSpinColourVectorD *)Grid.ShmBufferMalloc(bytes);
rbuf[d] = (HalfSpinColourVectorD *)Grid.ShmBufferMalloc(bytes);
bzero((void *)xbuf[d],bytes);
bzero((void *)rbuf[d],bytes);
}
int ncomm;
int bytes=lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD);
double dbytes;
for(int i=0;i<Nloop;i++){
double start=usecond();

5
documentation/GridXcode/readme.md
View File

@ -262,7 +262,6 @@ Set HEADER_SEARCH_PATHS to:
$Grid/build$(CONFIGURATION)/Grid
$Grid
$Grid/Grid
followed by (***the order is important***) the locations reported by `grid-config --cxxflags`, ignoring duplicates, e.g.:
@ -272,7 +271,7 @@ followed by (***the order is important***) the locations reported by `grid-confi
**Note: the easiest way to set this value is to put it all on one line, space separated, and edit the text to the right of `HEADER_SEARCH_PATHS`**, i.e.:
$Grid/build$(CONFIGURATION)/Grid $Grid $Grid/Grid $GridPre/openmpi/include $GridPkg/include $GridPre/lime/include
$Grid/build$(CONFIGURATION)/Grid $Grid $GridPre/openmpi/include $GridPkg/include $GridPre/lime/include
#### LIBRARY_SEARCH_PATHS
@ -298,7 +297,7 @@ The easiest way to link to all required libraries is to obtain a list of all lib
and pasting the output ***with `-lGrid -lHadrons ` prepended*** (including the `-l` switches) directly into `OTHER_LDFLAGS`, e.g.:
-lGrid -lHadrons -lmpi -lhdf5_cpp -lz -lcrypto -llime -lfftw3f -lfftw3 -lmpfr -lgmp -lstdc++ -lm -lz -lhdf5
-lGrid -lHadrons -lmpi -lhdf5_cpp -lhdf5 -lz -lcrypto -llime -lfftw3f -lfftw3 -lmpfr -lgmp -lm
## Make additional configurations

2
tests/forces/Test_dwf_force_eofa.cc
View File

@ -84,7 +84,7 @@ int main (int argc, char** argv)
DomainWallEOFAFermionR Rop(U, *FGrid, *FrbGrid, *UGrid, *UrbGrid, mb, mf, mb, -1.0, 1, M5);
OneFlavourRationalParams Params(0.95, 100.0, 5000, 1.0e-12, 12);
ConjugateGradient<LatticeFermion> CG(1.0e-12, 5000);
ExactOneFlavourRatioPseudoFermionAction<WilsonImplR> Meofa(Lop, Rop, CG, Params, true);
ExactOneFlavourRatioPseudoFermionAction<WilsonImplR> Meofa(Lop, Rop, CG, CG, CG, CG, CG, Params, true);
Meofa.refresh(U, RNG5);
RealD S = Meofa.S(U); // pdag M p

2
tests/forces/Test_dwf_gpforce_eofa.cc
View File

@ -89,7 +89,7 @@ int main (int argc, char** argv)
FermionAction Rop(U, *FGrid, *FrbGrid, *UGrid, *UrbGrid, mb, mf, mb, -1.0, 1, M5, params);
OneFlavourRationalParams Params(0.95, 100.0, 5000, 1.0e-12, 12);
ConjugateGradient<FermionField> CG(1.0e-12, 5000);
ExactOneFlavourRatioPseudoFermionAction<FermionImplPolicy> Meofa(Lop, Rop, CG, Params, true);
ExactOneFlavourRatioPseudoFermionAction<FermionImplPolicy> Meofa(Lop, Rop, CG, CG, CG, CG, CG, Params, true);
Meofa.refresh(U, RNG5);
RealD S = Meofa.S(U); // pdag M p

2
tests/forces/Test_mobius_force_eofa.cc
View File

@ -86,7 +86,7 @@ int main (int argc, char** argv)
MobiusEOFAFermionR Rop(U, *FGrid, *FrbGrid, *UGrid, *UrbGrid, mb, mf, mb, -1.0, 1, M5, b, c);
OneFlavourRationalParams Params(0.95, 100.0, 5000, 1.0e-12, 12);
ConjugateGradient<LatticeFermion> CG(1.0e-12, 5000);
ExactOneFlavourRatioPseudoFermionAction<WilsonImplR> Meofa(Lop, Rop, CG, Params, false);
ExactOneFlavourRatioPseudoFermionAction<WilsonImplR> Meofa(Lop, Rop, CG, CG, CG, CG, CG, Params, false);
Meofa.refresh(U, RNG5);
RealD S = Meofa.S(U); // pdag M p

2
tests/forces/Test_mobius_gpforce_eofa.cc
View File

@ -91,7 +91,7 @@ int main (int argc, char** argv)
FermionAction Rop(U, *FGrid, *FrbGrid, *UGrid, *UrbGrid, mb, mf, mb, -1.0, 1, M5, b, c, params);
OneFlavourRationalParams Params(0.95, 100.0, 5000, 1.0e-12, 12);
ConjugateGradient<FermionField> CG(1.0e-12, 5000);
ExactOneFlavourRatioPseudoFermionAction<FermionImplPolicy> Meofa(Lop, Rop, CG, Params, false);
ExactOneFlavourRatioPseudoFermionAction<FermionImplPolicy> Meofa(Lop, Rop, CG, CG, CG, CG, CG, Params, false);
Meofa.refresh(U, RNG5);
RealD S = Meofa.S(U); // pdag M p

10
tests/forces/Test_rect_force.cc
View File

@ -73,7 +73,7 @@ int main (int argc, char ** argv)
////////////////////////////////////
// Modify the gauge field a little
////////////////////////////////////
RealD dt = 0.0001;
RealD dt = 0.002;
LatticeColourMatrix mommu(&Grid);
LatticeColourMatrix forcemu(&Grid);
@ -88,13 +88,7 @@ int main (int argc, char ** argv)
// fourth order exponential approx
parallel_for(auto i=mom.begin();i<mom.end();i++){ // exp(pmu dt) * Umu
Uprime[i](mu) = U[i](mu) + mom[i](mu)*U[i](mu)*dt
+ mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt/2.0)
+ mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt/6.0)
+ mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt*dt/24.0)
+ mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt*dt*dt/120.0)
+ mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt*dt*dt*dt/720.0);
Uprime[i](mu) = U[i](mu) + mom[i](mu)*U[i](mu)*dt ;
}
}