1
0
mirror of https://github.com/paboyle/Grid.git synced 2024-09-20 01:05:38 +01:00
Go to file
Peter Boyle 2de009d994 Patch
2015-04-03 05:33:13 +01:00
m4 Build progressing 2015-03-04 04:34:51 +00:00
.gitignore Updating build system 2015-03-04 04:53:40 +00:00
aclocal.m4 MPI is now working and passing basic tests. Will start to construct a more sensible test suite shortly 2015-04-03 04:52:53 +01:00
AUTHORS Update AUTHORS 2015-03-07 07:00:39 +00:00
ChangeLog Updating build system 2015-03-04 04:53:40 +00:00
compile files 2015-03-04 11:57:14 +00:00
configure MPI is now working and passing basic tests. Will start to construct a more sensible test suite shortly 2015-04-03 04:52:53 +01:00
configure.ac MPI is now working and passing basic tests. Will start to construct a more sensible test suite shortly 2015-04-03 04:52:53 +01:00
COPYING Extra files 2015-03-04 12:03:07 +00:00
depcomp files 2015-03-04 11:57:14 +00:00
Grid_aligned_allocator.h Renamed the namespace to Grid 2015-04-03 05:29:54 +01:00
Grid_Cartesian.h Renamed the namespace to Grid 2015-04-03 05:29:54 +01:00
Grid_Communicator.h Renamed the namespace to Grid 2015-04-03 05:29:54 +01:00
Grid_config.h Merge branch 'master' of https://github.com/paboyle/Grid 2015-04-03 05:32:12 +01:00
Grid_config.h.in MPI is now working and passing basic tests. Will start to construct a more sensible test suite shortly 2015-04-03 04:52:53 +01:00
Grid_cshift_common.h MPI is now working and passing basic tests. Will start to construct a more sensible test suite shortly 2015-04-03 04:52:53 +01:00
Grid_cshift_fake.h Rename some files to make naming consistent 2015-04-03 04:58:03 +01:00
Grid_cshift_mpi.h Rename some files to make naming consistent 2015-04-03 04:58:03 +01:00
Grid_cshift_none.h Patch 2015-04-03 05:33:13 +01:00
Grid_cshift.h Rename some files to make naming consistent 2015-04-03 04:58:03 +01:00
Grid_fake.cc Renamed the namespace to Grid 2015-04-03 05:29:54 +01:00
Grid_init.cc Renamed the namespace to Grid 2015-04-03 05:29:54 +01:00
Grid_Lattice.h Renamed the namespace to Grid 2015-04-03 05:29:54 +01:00
Grid_main.cc Renamed the namespace to Grid 2015-04-03 05:29:54 +01:00
Grid_math_types.h Renamed the namespace to Grid 2015-04-03 05:29:54 +01:00
Grid_mpi.cc Renamed the namespace to Grid 2015-04-03 05:29:54 +01:00
Grid_QCD.h Renamed the namespace to Grid 2015-04-03 05:29:54 +01:00
Grid_simd.h Renamed the namespace to Grid 2015-04-03 05:29:54 +01:00
Grid_vComplexD.h Renamed the namespace to Grid 2015-04-03 05:29:54 +01:00
Grid_vComplexF.h Renamed the namespace to Grid 2015-04-03 05:29:54 +01:00
Grid_vRealD.h Renamed the namespace to Grid 2015-04-03 05:29:54 +01:00
Grid_vRealF.h Renamed the namespace to Grid 2015-04-03 05:29:54 +01:00
Grid.h Renamed the namespace to Grid 2015-04-03 05:29:54 +01:00
INSTALL Update INSTALL 2015-03-07 07:09:09 +00:00
install-sh file 2015-03-04 11:55:44 +00:00
LICENSE Initial commit 2015-03-04 02:30:11 +00:00
Makefile.am MPI is now working and passing basic tests. Will start to construct a more sensible test suite shortly 2015-04-03 04:52:53 +01:00
Makefile.in MPI is now working and passing basic tests. Will start to construct a more sensible test suite shortly 2015-04-03 04:52:53 +01:00
missing files 2015-03-04 11:57:14 +00:00
NEWS Updating build system 2015-03-04 04:53:40 +00:00
README Update README 2015-03-07 07:19:01 +00:00
README.md Update README.md 2015-03-07 07:20:12 +00:00

Grid

Data parallel C++ mathematical object library

This library provides data parallel C++ container classes with internal memory layout that is transformed to map efficiently to SIMD architectures. CSHIFT facilities are provided, similar to HPF and cmfortran, and user control is given over the mapping of array indices to both MPI tasks and SIMD processing elements.

  • Identically shaped arrays then be processed with perfect data parallelisation.
  • Such identically shapped arrays are called conformable arrays.

The transformation is based on the observation that Cartesian array processing involves identical processing to be performed on different regions of the Cartesian array.

The library will (eventually) both geometrically decompose into MPI tasks and across SIMD lanes.

Data parallel array operations can then be specified with a SINGLE data parallel paradigm, but optimally use MPI, OpenMP and SIMD parallelism under the hood. This is a significant simplification for most programmers.

The layout transformations are parametrised by the SIMD vector length. This adapts according to the architecture. Presently SSE2 (128 bit) AVX, AVX2 (256 bit) and IMCI and AVX512 (512 bit) targets are supported.

These are presented as

vRealF, vRealD, vComplexF, vComplexD

internal vector data types. These may be useful in themselves for other programmers. The corresponding scalar types are named

RealF, RealD, ComplexF, ComplexD

MPI parallelism is UNIMPLEMENTED and for now only OpenMP and SIMD parallelism is present in the library.

You can give `configure' initial values for configuration parameters by setting variables in the command line or in the environment. Here is are examples:

 ./configure CXX=clang++ CXXFLAGS="-std=c++11 -O3 -mavx" --enable-simd=AVX1

 ./configure CXX=clang++ CXXFLAGS="-std=c++11 -O3 -mavx2" --enable-simd=AVX2

 ./configure CXX=icpc CXXFLAGS="-std=c++11 -O3 -mmic" --enable-simd=AVX512 --host=none