ICODE-ADC/lammps
4
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

git-svn-id: svn://svn.icms.temple.edu/lammps-ro/trunk@12543 f3b2605a-c512-4ea7-a41b-209d697bcdaa

Browse Source
This commit is contained in:
sjplimp
2014-09-25 15:48:48 +00:00
parent dda702c5f9
commit 0dd4b6580f
1 changed files with 43 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files

61
src/MAKE/README
View File

@ -35,42 +35,67 @@ dir for the same thing using the Intel icc compiler instead of g++.
The OPTIONS directory has several Makefiles that extend either
Makefile.mpi or Makefile.serial to enable use of different compilers,
different MPIs, or several package and command options that LAMMPS
supports.
different MPIs, or package and command options that LAMMPS supports.
Makefile.fftw FFTW support added for long-range Coulombics (PPPM)
Makefile.jpg JPG support added to Makefile.mpi for dump image
Makefile.png PNG support added to Makefile.mpi for dump image
These Makefiles are the same as Makefile.mpi except they each add
settings for commonly used options (FFTW, JPG, PNG) to illustrate how
they can be added to any Makefile.
--------
Makefile.serial_icc = build without MPI for serial execution (single core)
Makefile.kokkos_omp KOKKOS package with CUDA support via nvcc compiler
Makefile.kokkos_cuda KOKKOS package with CUDA support via nvcc compiler
Makefile.kokkos_phi KOKKOS package with CUDA support via nvcc compiler
The fftw, jpg, and png Makefiles are the same as Makefile.mpi
except they each add settings for commonly used options (FFTW, JPG, PNG)
to illustrate how they can be added to any Makefile.
These Makefiles are for building with the KOKKOS package for different
architectures (multi-core with OpenMP threading, GPU, Xeon Phi). Some
Makefile variables are set and appended to.
--------
What about OPTIONS and MACHINES and CUSTOM/USER
Makefile.intel USER-INTEL package with CPU acceleration
Makefile.intel_offload USER-INTEL package with Xeon Phi support
These Makefiles are for building with the USER-INTEL package with
either CPU or Xeon Phi (offload) support. Some Makefile variables are
set and appended to.
--------
Makefile.fftw = add FFTW support to Makefile.mpi for long-range Coulombics
Makefile.jpg = add JPG support to Makefile.mpi for dump image command
Makefile.png = add PNG support to Makefile.mpi for dump image command
Makefile.mpich_g++ compile with MPICH wrapper, set compiler to GNU g++
Makefile.mpich_icc compile with MPICH wrapper, set compiler to Intel icc
Makefile.ompi_g++ compile with OpenMPI wrapper, set compiler to GNU g++
Makefile.ompi_icc compile with OpenMPI wrapper, set compiler to Intel icc
Makefile.intel
Makefile.intel_offload
Makefile.kokkos_omp
Makefile.kokkos_cuda
Makefile.kokkos_phi
Makefile.serial_icc serial build with Intel icc compiler
These Makefiles adapt Makefile.mpi and Makefile.serial to
use specific compilers.
--------
Makefile.mpich_native_g++ compile with GNU g++, link to MPICH
Makefile.mpich_native_icc compile with Intel icc, link to MPICH
Makefile.ompi_native_g++ compile with GNU g++, link to OpenMPI
Makefile.ompi_native_icc compile with Intel icc, link to OpenMPI
These Makefiles show how to build with MPI support, but not using the
MPI wrapper "mpicxx" for compiling. Instead, g++ or icc is used
directly to compile/link and the MPI library is linked to. This
requires knowing the path for the mpi.h and MPI library files.
------------------------------------
The MACHINE sub-directory has many Makefiles, each of which has been
setup for a specific kind of machine or a specific supercomputer.
Examples include
Examples include:
Makefile.mac Macintosh
Makefile.mac Apple Macintosh
Makefile.xe6 Cray XE6
Makefile.bgp IBM BG/Q
Makefile.bgq IBM BG/Q
Makefile.stampede U Texas "stampede" machine