c96ac858bf
GPU Package: Adding JIT test for OpenCL prefetch support.
2023-03-07 21:43:19 -08:00
37f22c8627
Misc Improvements to GPU Package
...
- Optimizations for molecular systems
- Improved kernel performance and greater CPU overlap
- Reduced GPU to CPU communications for discrete devices
- Switch classic Intel makefiles to use LLVM-based compilers
- Prefetch optimizations supported for OpenCL
- Optimized data repack for quaternions
2023-03-05 21:03:12 -08:00
2ccfe635ce
Removed the outdated CUDA_PROXY flag, using CUDA_MPS_SUPPORT consistently in CMake and traditional builds
2023-03-01 16:38:50 -06:00
4244d2e6cd
silence compiler warnings about unused parameters and variables
2023-01-19 08:56:54 -05:00
eddd3d6f25
Fixed a bug with extra being nullptr when _host_view is true: always allocate extra
...
(Note that BaseAmoeba has its own cast_extra_data() that doesn't know if extra is allocated properly, it is the case when _host_view is false for dedicated GPUs for example)
2023-01-18 20:04:45 -06:00
9dc0369cee
Attempted to resolve the address space change issue when casting for OpenCL 2.0 (ref: https://www.intel.com/content/www/us/en/developer/articles/technical/the-generic-address-space-in-opencl-20.html#06_address_space_casting )
2023-01-15 23:28:48 -06:00
8af77c690c
Merge branch 'develop' into amoeba-gpu
2022-12-14 13:16:41 -06:00
80a141d9c8
silence compiler warnings
2022-11-01 03:38:08 -04:00
9d081a5916
more adjustments for bogus timer results on Intel OpenCL
2022-10-19 07:39:56 -04:00
f867adc541
GPU Package fix where timing disable could result in event/marker destruction before completion on accelerator during initialization.
2022-10-19 02:16:29 -04:00
00f46120c7
Removed max_cus() from Device, used device->gpu->cus() instead
2022-10-07 15:50:30 -05:00
5a98a38e24
GPU Package: Switching to parallel GPU initialization / JIT compilation.
2022-10-07 13:25:14 -07:00
f715f174bb
GPU Package: Print OCL platform name to screen when multiple platforms
2022-10-06 21:40:42 -07:00
6b9e83fe20
Added timing for the induced dipole spreading part, computed the block size to ensure all the CUs are occupied by the fphi_uind and fphi_mpole kernels
2022-10-06 15:03:58 -05:00
2ef6a59c0a
Merge branch 'develop' into amoeba-gpu
2022-10-01 00:38:24 -05:00
6e34d21b24
GPU Package: Switching back to timer disabling with multiple MPI tasks per GPU. Logic added to prevent mem leak.
2022-09-28 21:02:16 -07:00
a14f0cfd6c
Merge branch 'amoeba' into amoeba-gpu, update the gpu pair styles with the base class
2022-06-28 12:54:27 -05:00
d4ea5ca49e
more clang-tidy fixes after re-running it with added settings
2022-05-14 07:18:05 -04:00
f09556018b
fix bugs reported by @jibril-b-coulibaly
2022-04-28 14:47:53 -04:00
531e553162
Merge branch 'amoeba' into amoeba-gpu
2022-04-22 16:10:24 -05:00
d6f7570d57
avoid redundant use of boolean literals
2022-04-10 20:47:31 -04:00
39b316729b
use auto type when assigning from cast or using new
2022-04-10 18:16:36 -04:00
a17bdf5652
silence compiler warnings and avoid infinite recursion in aspherical pair styles
2022-02-11 21:06:16 -05:00
87b63f768f
Only check for GPU double precision support if a GPU is present
2021-10-18 12:15:05 -04:00
afad3f42d5
Report only compatible GPU, i.e. no GPU if mixed/double precision is requested by the hardware does not support it
2021-10-13 21:15:16 -04:00
42dca75225
add check and suitable error message when fp64 is required but not available
2021-09-24 12:17:58 -04:00
17ba0d5804
possible workaround for some GPU package neighbor list issue
2021-09-22 21:47:32 -04:00
e6fb0e3bd8
small tweaks
2021-09-17 16:51:37 -04:00
353b3a2bb3
reformat for increased readability
2021-09-16 07:25:04 -04:00
2de482f825
Merge pull request #2911 from akohlmey/fix-gpu-package-issues
...
Fix minor GPU package issues for the stable release
2021-08-30 13:45:23 -04:00
39d8b239ff
don't report bogus timings
2021-08-29 17:56:47 -04:00
89556f0bcb
Override any OpenCL fast math JIT settings for born/coul/wolf{/cs}/gpu to resolve numerical deviations seen with some OpenCL implementations.
2021-08-28 17:01:58 -07:00
91317b2879
Added changes to Atom and Device classes for allocation of extra fields and SBBITS15 and NEIGHMASK15
2021-08-26 09:33:20 -05:00
a687868c69
finalize available GPU hardware introspection functions
2021-05-10 16:34:27 -04:00
fbdcfb2f72
preliminary interface to detect whether a viable GPU is present
2021-05-10 09:16:51 -04:00
299ad3b37d
work around bogus device overhead info in OpenCL
2021-05-08 23:43:15 -04:00
717faa6515
correctly detect GPU package with CUDA api
2021-04-06 19:12:28 -04:00
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
e7e2d2323b
Feb2021 GPU Package Update - GPU Package Files
2021-02-15 08:20:50 -08:00
56909e88b1
implement accelerator introspection for GPU package
2021-01-11 17:03:23 -05:00
f1ef7d85a8
T2345: Replace instances of NULL with nullptr
...
The following changes have been applied to src and lib folders:
regex replace: ([^"_])NULL ⇒ \1nullptr (8968 chgs in src, 1153 in lib)
Manually find/change: (void \*) nullptr ⇒ nullptr (1 case)
regex find: ".*?nullptr.*?"
Manually ~14 cases back to "NULL" in src, ~2 in lib
regex finds a few false positive where nullptr appears between two
strings in a function call
2020-09-12 09:34:38 -06:00
66c5fa2abd
Merge 'gpu_hip_port' into master
2020-01-28 20:35:08 +03:00
e832b5d50b
make clang++ happy when trying to compile the GPU library
2019-07-12 15:42:16 -04:00
8c3d18520d
add missing include needed on ppc64le
2019-06-26 10:45:31 -06:00
4a4297591e
Did some more cleanups
2019-04-17 12:04:31 -05:00
1f43efc111
Cleaned up the changes in Device and the base class of the pair styles
2019-04-17 00:09:49 -05:00
c55009a0ac
Enabled neighbor list build on the device with pair_style hybrid and hybrid/overlay
2019-04-16 23:30:25 -05:00
cd6b23d104
explicitly request OpenCL version 1.2 compatibility when compiling GPU package kernels for OpenCL
2019-03-22 09:50:31 -04:00
923ae041dc
(temporary) workaround for memory leaks with OpenCL and MPI for upcoming stable release
2018-07-23 15:52:42 -04:00
de8176b4fc
various minor OpenCL related fixes and improvements to the GPU package
...
- document previously undocumented OpenCL tune settings
- implement OpenCL platform selection through prefixing the device type with the platform id separated by a colon
- allow passing custom tune parameters though postfixing the device type with the 13 tuneable parameters separated by commas
- remove an extra clear() that would delete device properties structs an cause LAMMPS to output garbage strings
2018-07-20 14:41:54 -04:00
