1
0
mirror of https://github.com/paboyle/Grid.git synced 2025-06-22 09:42:02 +01:00

Merge GPU support (upstream/develop) into distillation branch.

This compiles and looks right ... but may need some testing

* develop: (762 commits)
  Tensor ambiguous fix
  Fix for GCC preprocessor/pragma handling bug
  Trips up NVCC for reasons I dont understand on summit
  Fix GCC complaint
  Zero() change
  Force a couple of things to compile on NVCC
  Remove debug code
  nvcc error suppress
  Merge develop
  Reduction finished and hopefully fixes CI regression fail on single precisoin and force
  Double precision variants for summation accuracy
  Update todo list
  Freeze the seed
  Fix compiling of MSource::Gauss for single precision
  Think the reduction is now sorted and cleaned up
  Fix force term
  Printing improvement
  GPU reduction fix and also exit backtrace option
  GPU friendly
  Simplify the comms benchmark
  ...

# Conflicts:
#	Grid/communicator/SharedMemoryMPI.cc
#	Grid/qcd/action/fermion/WilsonKernelsAsm.cc
#	Grid/qcd/action/fermion/implementation/StaggeredKernelsAsm.h
#	Grid/qcd/smearing/StoutSmearing.h
#	Hadrons/Modules.hpp
#	Hadrons/Utilities/Contractor.cc
#	Hadrons/modules.inc
#	tests/forces/Test_dwf_force_eofa.cc
#	tests/forces/Test_dwf_gpforce_eofa.cc
This commit is contained in:
Michael Marshall
2019-09-13 13:30:00 +01:00
796 changed files with 41536 additions and 52391 deletions

View File

@ -31,11 +31,9 @@
@file StoutSmearing.h
@brief Declares Stout smearing class
*/
#ifndef STOUT_SMEAR_
#define STOUT_SMEAR_
#pragma once
namespace Grid {
namespace QCD {
NAMESPACE_BEGIN(Grid);
/*! @brief Stout smearing of link variable. */
template <class Gimpl>
@ -90,8 +88,8 @@ public:
~Smear_Stout() {} // delete SmearBase...
void smear(GaugeField& u_smr, const GaugeField& U) const {
GaugeField C(U._grid);
GaugeLinkField tmp(U._grid), iq_mu(U._grid), Umu(U._grid);
GaugeField C(U.Grid());
GaugeLinkField tmp(U.Grid()), iq_mu(U.Grid()), Umu(U.Grid());
std::cout << GridLogDebug << "Stout smearing started" << std::endl;
@ -136,7 +134,7 @@ public:
// the i sign is coming from outside
// input matrix is anti-hermitian NOT hermitian
GridBase* grid = iQ._grid;
GridBase* grid = iQ.Grid();
GaugeLinkField unity(grid);
unity = 1.0;
@ -159,7 +157,7 @@ public:
Complex one_over_three = 1.0 / 3.0;
Complex one_over_two = 1.0 / 2.0;
GridBase* grid = u._grid;
GridBase* grid = u.Grid();
LatticeComplex c0(grid), c1(grid), tmp(grid), c0max(grid), theta(grid);
// sign in c0 from the conventions on the Ta
@ -178,7 +176,7 @@ public:
void set_fj(LatticeComplex& f0, LatticeComplex& f1, LatticeComplex& f2,
const LatticeComplex& u, const LatticeComplex& w) const {
GridBase* grid = u._grid;
GridBase* grid = u.Grid();
LatticeComplex xi0(grid), u2(grid), w2(grid), cosw(grid);
LatticeComplex fden(grid);
LatticeComplex h0(grid), h1(grid), h2(grid);
@ -219,7 +217,5 @@ public:
return cos(w) / (w * w) - sin(w) / (w * w * w);
}
};
}
}
#endif
NAMESPACE_END(Grid);