Ye Luo
e41530d353
Default CUDA host compiler as CXX compiler.
...
1) Require CUDA 11
2) -allow-unsupported-compiler by default.
2022-04-22 17:26:57 -05:00
Ye Luo
d2304738d9
Lift CUDA/HIP standard to C++17.
2022-04-19 11:12:22 -05:00
Ye Luo
4afb46ccfd
Remove ROCM/AOMP workaround.
2022-04-11 21:56:23 -05:00
Paul R. C. Kent
8ef5232226
Merge branch 'develop' into sycl-runtime
2022-04-07 19:49:28 -04:00
Paul R. C. Kent
7e2124b801
Set develop v3.14.9
2022-04-06 17:23:32 -04:00
Paul Kent
801bead310
3.14.0 version number and release notes
2022-04-06 14:25:25 -04:00
Ye Luo
c79a7f1998
Add partial OpenMP interop.
2022-03-30 17:35:25 -05:00
Ye Luo
dd97174884
Connect SYCL runtime.
2022-03-30 17:35:25 -05:00
Ye Luo
d422f3238b
Refine macro usage.
...
ENABLE_ROCM is a CMake option not a macro. It should not be used in the source codes.
2022-03-25 13:40:18 -05:00
Ye Luo
5003cad645
Format cmake files.
2022-03-25 10:36:07 -05:00
Ye Luo
a6308e3554
Minor cmake cleaning.
2022-02-24 15:56:24 -06:00
Paul Kent
53038ad50e
Set dev version 3.13.9
2022-02-16 09:12:07 -05:00
Paul Kent
7ef61ab9d1
Increase version number
2022-02-14 10:00:45 -05:00
Paul R. C. Kent
d27e81818c
Merge branch 'develop' into gpaw_converter
2022-02-08 16:43:56 -05:00
Ye Luo
1d3222430e
Remove USE_REAL_STRUCT_FACTOR
2022-02-06 13:28:08 -06:00
Juha Tiihonen
dd45cbc98a
Fix python requirements for GPAW tests
2022-01-28 08:43:36 +02:00
Juha Tiihonen
90f620d976
Merge branch 'develop' of https://github.com/QMCPACK/qmcpack into gpaw_converter
2022-01-25 09:55:22 +02:00
William F Godoy
8e55b9a442
Remove ENABLE_OPENMP in favor of QMC_OMP
...
CMake configuration files
Source code pre-processor
2022-01-20 14:30:00 -05:00
Steven Hahn
51f8281d66
Use check_python_reqs
...
Signed-off-by: Steven Hahn <hahnse@ornl.gov>
2022-01-14 17:51:03 -05:00
Steven Hahn
55b1df65aa
Use standard check for python library instead of HAVE_PYSCF
...
Signed-off-by: Steven Hahn <hahnse@ornl.gov>
2022-01-14 16:15:04 -05:00
Ye Luo
4595d71a3c
Fix nexus install path.
2022-01-11 20:20:37 -06:00
Ye Luo
f4d4fc3bc8
Remove BoostRandom
2022-01-07 11:35:32 -06:00
Ye Luo
287a550084
Merge branch 'develop' into add-qmcrng
2022-01-06 20:01:02 -06:00
Ye Luo
4587bf5a8b
Introduce qmc_external_XXX targets.
2022-01-06 09:26:05 -06:00
Juha Tiihonen
9d8be6d19a
Merge branch 'develop' of https://github.com/QMCPACK/qmcpack into gpaw_converter
2022-01-04 12:58:24 +02:00
Ye Luo
7982a725ec
Set QMC_RNG_BOOST=OFF by default.
2022-01-03 16:36:00 -06:00
Ye Luo
55aec4fd0b
Add QMC_RNG_BOOST option.
2022-01-01 14:04:47 -06:00
Steven Hahn
d82a7f5c40
Add timestamp to cache name
...
Signed-off-by: Steven Hahn <hahnse@ornl.gov>
2021-12-20 17:36:55 -05:00
Juha Tiihonen
09e0e04807
Relocate GPAW converter tests, refine build conditions
2021-12-20 12:08:23 +02:00
Juha Tiihonen
198d6a98a6
Add CMake dependencies for GPAW4QMCPACK converter
2021-12-20 10:46:58 +02:00
Ye Luo
39a9bd921a
CMake fix for NVHPC 21.11
2021-12-11 21:44:06 -06:00
Paul R. C. Kent
36f4498881
Set development version after release to 3.12.9
2021-12-08 16:02:29 -05:00
Paul Kent
aea2109be6
v3.12.0 release notes and version number
2021-12-07 14:21:00 -05:00
Kevin Gasperich
2c533a4d1b
add real-space RMG converter to convert4qmc ( #3351 )
...
* started adding RMG to convert4qmc
* replace raw pointer with std::unique_ptr for RMGParser in convert4qmc
* clang-format
* updated docs for RMG converter
* change supertwist handling in RMG converter
* added tests for RMG converter
* removed larger rmg converter test
* added spin-polarized RMG converter test
* add override for two RMGParser functions
* added RMG ctest (WIP)
* cleaned up RMG tests
* shortened rmg vmc test
* added rmg-diamond reference vals
* minor cleanup
* minor cleanup of cmake for RMG
Co-authored-by: Paul R. C. Kent <kentpr@ornl.gov>
2021-11-03 21:27:16 +00:00
Ye Luo
a9e7f91273
Test compile main after compile option setting.
2021-11-03 11:29:36 -05:00
Ye Luo
a4e777e674
Turn on ROCM workaround by default.
2021-11-02 12:00:45 -05:00
Ye Luo
21809500cc
Set QMC_OFFLOAD_MEM_ASSOCIATED automatically for ROCM.
2021-11-02 10:49:14 -05:00
Ye Luo
f3ead65dd2
More offload flag automation for AMD.
2021-11-02 10:36:04 -05:00
Ye Luo
7592098910
Merge branch 'develop' into omp-use-external-memory
2021-11-01 21:09:42 -05:00
Ye Luo
beb95184d4
Add more warning and traps.
2021-11-01 15:35:02 -05:00
Ye Luo
fef6b26b3b
Introduce QMC_OFFLOAD_MEM_ASSOCIATED cmake option.
...
Some OpenMP runtime library implementations may not have good interoperability with vendor runtime systems.
The outstanding issue is that a device pointer allocated via OpenMP runtime library cannot be de-referenced in a kernel or vendor library created with the vendor native programming model.
If the reverse way works, namely a device pointer allocated by the vendor native programming model can be de-referenced in OpenMP offload region. We can always allocate memory from the vendor native programming model and associate it to OpenMP runtime.
2021-10-31 19:34:00 -05:00
Ye Luo
f989d122be
Harden ROCM in CMake.
2021-10-31 17:32:56 -05:00
Ye Luo
058e446518
Update recipes under config for CUDA change.
2021-09-30 10:06:18 -04:00
Ye Luo
b39aea6a78
Set CMAKE_CUDA_ARCHITECTURES early.
2021-09-29 19:06:57 -05:00
Ye Luo
57e55d33e2
Set back CUDA default to C++14.
2021-09-29 17:30:50 -05:00
Ye Luo
734415c99f
CMAKE_CUDA_ARCHITECTURES needs CMake 3.18.
2021-09-29 17:24:08 -05:00
Ye Luo
707977ed68
More accurate stopper message.
2021-09-29 17:22:44 -05:00
Steven Hahn
761745bfae
Apply changes recommended by @ye-lou.
...
Signed-off-by: Steven Hahn <hahnse@ornl.gov>
2021-09-29 17:36:10 -04:00
Steven Hahn
b607510cfc
Don't change required CMake version
...
Signed-off-by: Steven Hahn <hahnse@ornl.gov>
2021-09-29 15:06:19 -04:00
Steven Hahn
ed13bd8918
Eliminate depricated find_package(CUDA) from qmcpack
...
Replace it with first-class language support and find_package(CUDAToolkit)
Signed-off-by: Steven Hahn <hahnse@ornl.gov>
2021-09-29 13:40:01 -04:00