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

use '-pk kokkos gpu/direct on/off' consistently in comments and docs

Browse Source
This commit is contained in:
Axel Kohlmeyer
2018-08-09 04:16:22 -04:00
parent 01b8134746
commit dd9fed11dc
3 changed files with 17 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
doc/src/Speed_kokkos.txt
View File

@ -102,9 +102,9 @@ the case, especially when using pre-compiled MPI libraries provided by
a Linux distribution. This is not a problem when using only a single a Linux distribution. This is not a problem when using only a single
GPU and a single MPI rank on a desktop. When running with multiple GPU and a single MPI rank on a desktop. When running with multiple
MPI ranks, you may see segmentation faults without GPU-direct support. MPI ranks, you may see segmentation faults without GPU-direct support.
These can be avoided by adding the flags '-pk kokkos gpu/direct no' These can be avoided by adding the flags '-pk kokkos gpu/direct off'
to the LAMMPS command line or using "package kokkos gpu/direct no"_package.html to the LAMMPS command line or by using the command
in the input file. "package kokkos gpu/direct off"_package.html in the input file.
Use a C++11 compatible compiler and set KOKKOS_ARCH variable in Use a C++11 compatible compiler and set KOKKOS_ARCH variable in
/src/MAKE/OPTIONS/Makefile.kokkos_cuda_mpi for both GPU and CPU as /src/MAKE/OPTIONS/Makefile.kokkos_cuda_mpi for both GPU and CPU as

2
doc/src/package.txt
View File

@ -625,7 +625,7 @@ switch"_Section_start.html#start_6.
For the KOKKOS package, the option defaults neigh = full, neigh/qeq = For the KOKKOS package, the option defaults neigh = full, neigh/qeq =
full, newton = off, binsize = 0.0, and comm = device, gpu/direct = on. full, newton = off, binsize = 0.0, and comm = device, gpu/direct = on.
These settings are made automatically by the required "-k on" These settings are made automatically by the required "-k on"
"command-line switch"_Section_start.html#start_6. You can change them bu "command-line switch"_Section_start.html#start_6. You can change them by
using the package kokkos command in your input script or via the "-pk using the package kokkos command in your input script or via the "-pk
kokkos" "command-line switch"_Section_start.html#start_6. kokkos" "command-line switch"_Section_start.html#start_6.

6
src/KOKKOS/kokkos.cpp
View File

@ -156,11 +156,11 @@ KokkosLMP::KokkosLMP(LAMMPS *lmp, int narg, char **arg) : Pointers(lmp)
} else if (-1 == have_gpu_direct() ) { } else if (-1 == have_gpu_direct() ) {
error->warning(FLERR,"Kokkos with CUDA assumes GPU-direct is available," error->warning(FLERR,"Kokkos with CUDA assumes GPU-direct is available,"
" but cannot determine if this is the case\n try" " but cannot determine if this is the case\n try"
" '-pk kokkos gpu/direct no' when getting segmentation faults"); " '-pk kokkos gpu/direct off' when getting segmentation faults");
} else if ( 0 == have_gpu_direct() ) { } else if ( 0 == have_gpu_direct() ) {
error->warning(FLERR,"GPU-direct is NOT available, but some parts of " error->warning(FLERR,"GPU-direct is NOT available, but some parts of "
"Kokkos with CUDA require it by default\n try" "Kokkos with CUDA require it by default\n try"
" '-pk kokkos gpu/direct no' when getting segmentation faults"); " '-pk kokkos gpu/direct off' when getting segmentation faults");
} else { } else {
; // should never get here ; // should never get here
} }
@ -310,7 +310,7 @@ void KokkosLMP::accelerator(int narg, char **arg)
} else error->all(FLERR,"Illegal package kokkos command"); } else error->all(FLERR,"Illegal package kokkos command");
} }
// if "gpu/direct no" and "comm device", change to "comm host" // if "gpu/direct off" and "comm device", change to "comm host"
if (!gpu_direct) { if (!gpu_direct) {
if (exchange_comm_classic == 0 && exchange_comm_on_host == 0) if (exchange_comm_classic == 0 && exchange_comm_on_host == 0)