biod.pnpi.spb.ru / alexxy / gromacs.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 678804b) | patch
Improved CUDA non-bonded kernel performance
authorSzilard Pall <pall.szilard@gmail.com>
Fri, 28 Feb 2014 21:11:08 +0000 (22:11 +0100)
committerSzilárd Páll <pall.szilard@gmail.com>
Fri, 28 Feb 2014 21:14:37 +0000 (22:14 +0100)
commit17e38531c7b9eb15654be49cf9d453c2f32a84d9
tree7a58772695866a108b6792a9aefdcfae97ffc047tree | snapshot
parent678804b8684b2c64e9db98bc266113aaf8a79cb5commit | diff
Improved CUDA non-bonded kernel performance

Some old tweak which was supposed to improve performance had in fact
the opposite effect. Removing this tweak and with it eliminating
shared memory bank conflicts it caused improved performance by up
to 2.5% in the force-only CUDA kernel.

Change-Id: I7fcb24defed2c68627457522c39805afc83b3276
src/mdlib/nbnxn_cuda/nbnxn_cuda_kernel.cuh diff | blob | history
Local GROMACS mirror with custom stuff