| commit | 18e384d3e37503add2d7fa01f7642706e160d7fd | |
| author | Berk Hess <hess@kth.se> | |
| Mon, 16 May 2016 20:33:27 +0000 (16 22:33 +0200) | ||
| committer | Mark Abraham <mark.j.abraham@gmail.com> | |
| Wed, 22 Jun 2016 07:43:58 +0000 (22 09:43 +0200) | ||
| tree | ea954551ef4fc2caf59cf5142d0fb42a937e1483 | treesnapshot (tar.gz zip) |
| parent | 194af8d36b0f8138dc9bee514e3808d4f963ac99 | commitdiff |
Remove OpenMP overhead at high parallelization
Commit 6d98622d introduced OpenMP parallelization for for loops
clearing rvecs of increasing rvecs. For small numbers of atoms per
MPI rank this can increase the cost of the loop by up to a factor 10.
This change disables OpenMP parallelization at low atom count.
Change-Id: I0006526568bb387f91e0a373f7ef203b3809f2e7
Commit 6d98622d introduced OpenMP parallelization for for loops
clearing rvecs of increasing rvecs. For small numbers of atoms per
MPI rank this can increase the cost of the loop by up to a factor 10.
This change disables OpenMP parallelization at low atom count.
Change-Id: I0006526568bb387f91e0a373f7ef203b3809f2e7
| src/gromacs/ewald/pme-pp.cpp | diffblobblamehistory | |
| src/gromacs/mdlib/gmx_omp_nthreads.h | diffblobblamehistory | |
| src/gromacs/mdlib/sim_util.cpp | diffblobblamehistory |