Commit Graph

31 Commits

Author SHA1 Message Date
58bf89fda3 Changing erroneous BIO_PAIR in kernels to BLOCK_PAIR. Default threads per atom is now 1 for all but charged pair styles. 2011-04-15 20:57:03 -04:00
5ff6092ea1 Updating all preprocessor defines to be set by a kernel. Adding a threads
per atom option to fix gpu.
2011-04-13 13:57:56 -04:00
2759b11312 Adding fixes for OpenCL support and correct constant casting to pppm
kernels.
2011-04-04 14:27:33 -04:00
a955c52830 Consolidated block size defines into device header. 2011-04-04 13:41:31 -04:00
f08932d407 Switch pencil size in PPPM to arch dependent. Commiting for testing on
yona.
2011-04-04 12:22:24 -04:00
86f7f6cd42 Adding better error output to LAMMPS routines. 2011-04-03 11:51:26 -04:00
a3eca0b125 Adding stuff to support async charge spread for PPPM. 2011-03-29 13:23:12 -04:00
71ff7b8096 Updating all code to use full neighbor lists for all host forces
computations. Changed the load balancing scheme. Added timing estimates
for GPU and driver overhead, cpu idle time.
2011-03-28 13:18:39 -04:00
8f218cc11d Another update for yona testing. 2011-03-27 19:50:44 -04:00
092ee2eea8 Temporary commit for testing on yona. 2011-03-27 12:31:18 -04:00
f0263bf14c pppm/gpu now using all gpu kernels. 2011-03-25 11:49:09 -04:00
e32f0861da Fixing pppm/gpu to work with gpu pair styles. 2011-03-24 19:10:12 -04:00
a56bc868aa Fixing bug in GPU selection. 2011-03-15 16:00:05 -04:00
1fb6410921 Adding PPPM interp kernel calls and tests. 2011-03-14 21:39:41 -04:00
a3d6cff836 CODE USED TO BENCHMARK CHARGE SPREADING FOR PAPER. 2011-02-21 11:35:26 -05:00
e8da16ff23 Finishing laptop optimization of PPPM charge spread kernels. 2011-02-17 13:37:02 -05:00
3e25ee5338 Adding kernel for rho map; not currently working. 2011-02-07 20:15:06 -05:00
3e4a467b20 Adding working GPU particle map for PPPM. Reverting changes to
force.h/cpp.
2011-02-02 14:12:22 -05:00
343485d592 Adding pppm stuff and incorporating data transfer. 2011-01-31 15:06:22 -05:00
c1c9fd47de Hybrid working, but not fully tested. 2011-01-27 17:52:08 -05:00
b9c7e2bb96 Trying to get hybrid gpu styles to work... 2011-01-25 17:46:00 -05:00
6018dd02b1 Untested changes to support hybrid styles. 2011-01-24 15:58:04 -05:00
3324437866 Started updating to support multiple GPU styles. Incomplete - non-compilable. 2011-01-23 20:03:42 -05:00
0232e672eb Revert "Removing msm, openmp, and k-space stuff for LAMMPS patch."
This reverts commit fd33d29442.

Conflicts:

	src/GPU/Install.sh
2011-01-21 11:13:30 -05:00
fd33d29442 Removing msm, openmp, and k-space stuff for LAMMPS patch. 2011-01-11 12:56:45 -05:00
c1e53cf140 Final working charmm commit. 2010-12-30 19:52:32 -05:00
b58bd96444 Now using LAMMPS communicators. 2010-12-15 18:42:49 -05:00
aca0eefff1 More efficient memory usage when particle split>0 and < 1. 2010-12-15 16:49:37 -05:00
af4911b377 Adding MSM version of CMM coul/long/gpu style to library. 2010-12-12 18:17:11 -05:00
4366bfffd3 Getting rid of extra CR characters at ends of lines.
git-svn-id: svn://svn.icms.temple.edu/lammps-ro/trunk@5285 f3b2605a-c512-4ea7-a41b-209d697bcdaa
2010-11-23 19:52:03 +00:00
5a82c99485 Changes from Mike Brown.
git-svn-id: svn://svn.icms.temple.edu/lammps-ro/trunk@5277 f3b2605a-c512-4ea7-a41b-209d697bcdaa
2010-11-23 00:40:35 +00:00