b4a70880d9
Merge pull request #2745 from rbberger/gpu_updates
...
Updates to the GPU package
2021-04-30 19:32:44 -04:00
6029445ab5
Fix tagint datatype with LAMMPS_BIGBIG in OpenCL
...
The previous version was trying to use 16 byte integers.
2021-04-30 12:01:56 -04:00
375f8e7054
support loading OpenCL functions up to the OpenCL 3.0 standard
2021-04-30 09:28:01 -04:00
8dfcdc4b3c
Revert "activate workaround for problems with new GPU code neighbor lists"
...
This reverts commit a28d9a6e69
2021-04-29 19:24:28 -04:00
24314b2316
Formatting
2021-04-23 16:36:01 -04:00
8ba1b59d8d
Correct fprintf statement in debug code
2021-04-23 16:34:12 -04:00
0632922a9b
Explicitly check for subgroup support instead of CL version
2021-04-22 12:54:13 -04:00
0f1f49afa7
Add more output to ocl_get_devices
2021-04-22 12:52:30 -04:00
d405f2ec4b
Update defines to use old neighbor code for CUDA >= 11.2
2021-04-22 09:16:37 -04:00
bb7931c989
Fixing bugs in slow (non-shared memory) variant of lj/charmm/coul/charmm/gpu
2021-04-18 09:35:54 -07:00
3d5897c926
Fixes issue from Feb 2021 GPU package update for tersoff styles using multiple types.
2021-04-18 01:40:33 -07:00
18e5e42ce3
minor change
2021-04-18 04:30:59 +03:00
a91e904f34
minor changes
2021-04-17 14:56:16 +03:00
5d00fa7ec5
update constants for lj/cubic/gpu from CPU version
2021-04-07 14:00:14 -04:00
3e3f66adac
Merge branch 'master' into smooth-gpu
...
Added gpu acceleration for smooth pair_style.
2021-04-07 14:15:33 +03:00
717faa6515
correctly detect GPU package with CUDA api
2021-04-06 19:12:28 -04:00
4c2fb7a431
Porting to new default platform for AMD/HIP in ROCm 4.1
2021-04-01 09:43:13 -05:00
9e412bb7a6
Fixing bug with GPU neighboring when using builds supporting CUDPP.
...
...introduced in Feb 2021 GPU package update. Manifests when GPU library is built with -DUSE_CUDPP and -DLAL_USE_OLD_NEIGHBOR (latter forced with CUDA 11.2).
2021-03-31 11:42:39 -07:00
dca9cd9c1c
Kept the .cubin files so that they are only rebuilt whenever changes are made to the relevant source files
2021-03-29 13:46:16 -05:00
7662cfe28d
Merge /home/ndtrung/Codes/lammps-git into eam-gpu
2021-03-29 13:33:31 -05:00
8d18051232
acceleration for pair_style lj/smooth
2021-03-18 20:48:57 +03:00
d1b4af60a3
disable GPU binning via CUDPP by default and describe why
2021-03-16 11:58:47 -04:00
a28d9a6e69
activate workaround for problems with new GPU code neighbor lists
2021-03-11 22:06:10 -05:00
ca88f97a4b
added acceleration of lj/smooth on gpu
2021-03-12 01:40:52 +03:00
e32d059d26
lj/smooth/gpu
2021-03-06 23:43:25 +03:00
4960aeb3c8
error on line 194
2021-03-06 23:39:37 +03:00
d5c3e1786a
Fixing bugs on eam*/gpu for pair hybrid with neigh yes, where the gpu pair style eam is used for only a subset of the pair types. eam being the first substyle works correctly, but otherwise will give incorrect forces
2021-03-02 13:05:37 -06:00
70ca055113
Add workaround to compile with HIP, new neighbor code still needs to be ported
2021-02-23 11:28:05 -05:00
2a0b7a26c0
Add missing cq() getter
2021-02-23 11:26:07 -05:00
446a068159
Use hipGetDeviceProperties to avoid int overflows
2021-02-23 10:14:05 -05:00
d44af3256d
Fix lal_charmm.cu to support HIP compilation
2021-02-23 09:51:27 -05:00
45c782308c
Fixing issue from recent GPU package update with OMP_NUM_THREADS env being overridden in GPU library.
...
Fixing race condition with OpenMP for GPU styles using torque (missed in regression tests due to the first fix)
Documenting GPU package option for setting the number of threads (consistent with USER-INTEL and USER-OMP).
2021-02-18 21:08:18 -08:00
f367e66aba
documentation corrections, spelling fixes and updates
2021-02-17 18:47:35 -05:00
e575c5fa29
-fopenmp implies -fopenmp-simd
2021-02-17 18:30:02 -05:00
1e5a73c468
silence warnings when using default OpenCL headers. Pick OpenCL v2.1 as default.
2021-02-17 17:36:35 -05:00
721c6d96cc
Removing Makefile.opencl from lib/gpu
2021-02-16 00:36:37 -08:00
e7e2d2323b
Feb2021 GPU Package Update - GPU Package Files
2021-02-15 08:20:50 -08:00
06e484d3e0
make flags for supporting GPU archs consistent, add ampere consumer GPU arch
2021-02-07 15:30:55 -05:00
754a469a01
add variable for handling cuda-mps-server
2021-02-05 16:43:51 -05:00
4166235be6
synchronize settings with CMake support
2021-02-05 16:43:18 -05:00
7b87f45d1a
Merge pull request #2554 from akohlmey/accelerator-introspection
...
Add accelerator introspection functionality
2021-01-18 11:59:00 -05:00
efda7559c0
Merge branch 'master' into accelerator-introspection
2021-01-16 00:15:07 -05:00
8407145e71
remove obsolete makefiles from lib/gpu based on info from @ndtrung81
2021-01-15 14:51:29 -05:00
537019f91f
Made Nvidia.makefile and Opencl.makefile more compact, stored their older version
2021-01-13 22:52:01 -06:00
57db3631e5
Added Makefile.cuda for CUDA and Makefile.opencl for OpenCL, similar to what has been done in Makefile.hip for HIP
2021-01-13 22:50:32 -06:00
56909e88b1
implement accelerator introspection for GPU package
2021-01-11 17:03:23 -05:00
28641bcbc7
Merge pull request #2402 from ndtrung81/gpu-dpd
...
Allowed dpd/tstat/gpu and dpd/gpu to work together in pair style hybrid
2020-10-14 17:18:02 -04:00
67803fe17f
Minor cleanup for dpd/tstat in the GPU library
2020-09-28 11:41:39 -05:00
28812b1ea7
Merge pull request #2381 from akohlmey/collected-small-changes
...
Collected small changes
2020-09-24 16:17:28 -04:00
9df1bb4db0
remove obsolete makefiles. Makefile.xk7 is on death watch but Blue Waters still has such nodes
2020-09-22 17:47:14 -04:00
