zhangyiss/spack
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

petsc: specify number of processors used for test case (#8015)

Browse Source 
For Spectrum MPI, if -np is omitted, the default behavior is to
assign one process per process slot, where the default process slot
allocation is one per core. On systems with many cores, the number
of processes can exceed the size of the grid specified when the
testcase is run and the test case fails. This specifies a small
number of processes when running the test case to prevent failure.
This commit is contained in:
David Wootton 2018-05-10 13:49:18 -04:00 committed by scheibelp
parent f4a46f4fc3
commit 9c4522ecbd
1 changed files with 12 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
var/spack/repos/builtin/packages/petsc/package.py
View File

@ -296,6 +296,18 @@ def install(self, spec, prefix):
cc('ex50.c', '-I%s' % prefix.include, '-L%s' % prefix.lib,
'-lpetsc', '-lm', '-o', 'ex50')
run = Executable(join_path(spec['mpi'].prefix.bin, 'mpirun'))
# For Spectrum MPI, if -np is omitted, the default behavior is
# to assign one process per process slot, where the default
# process slot allocation is one per core. On systems with
# many cores, the number of processes can exceed the size of
# the grid specified when the testcase is run and the test case
# fails. Specify a small number of processes to prevent
# failure.
# For more information about Spectrum MPI invocation, see URL
# https://www.ibm.com/support/knowledgecenter/en/SSZTET_10.1.0/smpi02/smpi02_mpirun_options.html
if ('spectrum-mpi' in spec):
run.add_default_arg('-np')
run.add_default_arg('4')
run('ex50', '-da_grid_x', '4', '-da_grid_y', '4')
if 'superlu-dist' in spec:
run('ex50',