ae92477389
Fix full double precision GPU complex code.
...
Fix cuda_inverse complex double inversion.
Remove warning in CMakeList.txt.
2017-01-21 00:32:24 -06:00
6dd24932e1
Improved phase factor.
...
nvcc is not happy with thrust::complex as __shared__.
eliminate shared buffer and data race.
2017-01-20 23:58:13 -06:00
4fddf3b18b
Clean up TrialWaveFunction_CUDA.cpp SlaterDet.h.
2017-01-20 23:47:34 -06:00
8bb6b38e23
Clean up DiracDeterminantCUDA.h/cpp for cplx GPU.
...
Merge CudaValueType with CudaRealType.
Also fix the GPU DP complex issue.
2017-01-20 18:29:26 -06:00
d1b3c8e0b6
Output warning for non-mixed precision complex GPU builds that results will be wrong.
2017-01-20 09:22:55 -05:00
d7b5cd2b89
Fixed compilation errors for non-complex GPU code caused by compiler directives mismatches introduced by complex GPU code.
2017-01-20 09:03:42 -05:00
69acfc6205
Added Complex GPU code to current development branch.
2017-01-19 18:03:57 -05:00
90b3fd1546
Merge pull request #75 from markdewing/git_info
...
Record git repository information in the build
2017-01-19 11:08:23 -05:00
1fba9520cc
Move non-git source version info to subdirectory
...
In the case where source is not a git repository, an empty repository info
file is created at configure time. The git-rev.h file is placed in the src/
subdirectory. If the code to create it is in the top-level CMakeLists.txt,
the src/ directory does not yet exist and it fails to create the file.
Move the git-ver.h file creation code to the src/CMakeLists.txt.
2017-01-18 23:02:22 -06:00
65acfe3978
Merge pull request #74 from jtkrogel/dev
...
test PR 2, remove a space from nexus.py
2017-01-18 17:16:18 -05:00
c4ab3a2f4f
test commit 2, remove a space from nexus.py
2017-01-18 17:06:37 -05:00
b7afe0b0a1
Record git repository information in the build.
...
Use a CMake custom command that runs every build to pick up git repository
information.
The code was inspired and informed by this Stack Overflow discussion:
http://stackoverflow.com/questions/1435953/how-can-i-pass-git-sha1-to-compiler-as-definition-using-cmake
2017-01-18 16:04:08 -06:00
0fc43ce2f1
Merge pull request #71 from markdewing/bgq_timer
...
Update timer for BG/Q.
2017-01-18 16:43:27 -05:00
4cd15dd301
Merge pull request #72 from ye-luo/fix-build-with-CRAYMPI-flag
...
fix the building with -D_CRAYMPI on Titan.
2017-01-18 16:39:01 -05:00
50be208cde
Merge pull request #67 from ye-luo/add-PW-openBC-test
...
Add pw open bc test
2017-01-18 16:18:07 -05:00
5a2f6abdda
Merge pull request #68 from prckent/Add_QE_to_tests
...
Add qe to tests
2017-01-18 15:05:58 -06:00
3dede6f567
fix the building with -D_CRAYMPI on Titan.
...
_CRAYMPI flag was used on the Titan toolchains in the attic folder but not in the most recent build script.
No clue if someone checked whether this flag is needed or not. The titan weekly tests shows OK without this flag.
Anyway, fix the broken code path since the AFQMC merge.
2017-01-18 16:04:43 -05:00
5bacd48896
Update timer for BG/Q.
...
Read directly from the time base register for a low-overhead timer.
This assumes a 1.6 GHz processor clock.
2017-01-18 21:01:55 +00:00
3ba0a3e7fa
Merge pull request #70 from jtkrogel/dev
...
test PR, add a space to nexus readme
2017-01-18 15:54:04 -05:00
10183b59fa
test commit, add a space to nexus readme
2017-01-18 15:43:23 -05:00
062ff7c1be
Merge pull request #69 from naromero77/develop
...
Update hyperion nightly regression test script for github
2017-01-18 15:22:43 -05:00
5f40816d4c
Update hyperion nightly script (part3)
...
Switch over to primary QMCPACK repo.
2017-01-18 14:02:13 -06:00
ca1515d016
Update hyperion nightly script (part2)
...
Fix paths and use flash drive.
2017-01-18 13:51:14 -06:00
aaff9669d1
Update ctest script to pass QE_BIN and enable Quantum Espresso testing. Add force to patch script for when run on top of existing source. Add building and caching of QE to Intel2015 builds on oxygen
2017-01-18 14:43:10 -05:00
cac0d93883
Remove vestiges of autotools files.
2017-01-18 12:01:02 -06:00
7dd8b150f8
slight change in the wording.
2017-01-18 00:31:23 -06:00
010731a7e9
add atom O open BC noj test.
2017-01-18 00:11:20 -06:00
104397cd7a
Update hyperion regression test for github
...
Update regression test that runs on the KNL box at ALCF to use
GitHub instead of the SVN repo.
2017-01-17 23:04:30 -06:00
e5885a261a
adjust optimizer tests.
...
1, Remove parameters identical to the the default.
2, The number of samples was not properly set. Now *16. The testing time should not change significantly.
3, Use buffer on the quartic optimizer to speed it up.
2017-01-17 19:13:27 -06:00
24944d53ad
add the atomO openBC test starting from qe.
...
1, Due to the large file size by the vacuum, the wf h5 is generated by pw.x from a DFT run. Once the h5 is generated, the QMC tests can run independently.
2, Enhanced CMake/run_qe.cmake. when proc2=0, nscf step is skipped.
2017-01-17 15:25:24 -06:00
24b9784afd
Recover the short-monoO flux estimator test.
2017-01-17 15:04:08 -06:00
48273db414
update ANL cetus nightly tests to adopt github.
2017-01-17 14:00:28 -06:00
5149528ae9
Fix script name for resubmit
2017-01-17 12:32:25 -06:00
a09caa0513
Updated OLCF scripts. In practice also use _cpuz variant for complex build and to avoid time limits
2017-01-17 12:32:25 -06:00
826b86fbe6
Merge pull request #62 from markdewing/timer_fixed_sized_struct
...
Lower stack timer overhead using fixed-size key.
2017-01-16 22:17:51 -05:00
65d3c2b878
Remove redundant OpenMP pragmas in the timers.
...
These functions are called from regions of code protected by 'omp master'.
2017-01-16 18:08:25 -06:00
056d65f607
Lower stack timer overhead using fixed-size key.
...
The map for collecting the per-stack timer data used a string representation
of the stack as a key. Some of the overhead comes from managing these strings.
The new key assigns a small integer timer id to each timer and constructs
the stack in a fixed-size array.
An additional hack to increase performance treats a chunk of timers as a larger
data type for the comparisons needed by the map data structure.
Current settings use 8 bits for the timer id (giving up to 254 timers), and
two 64 bit ints for the stack (16 stack levels). If these are exceeded, warnings
are printed to the screen and in the .info.xml file.
Use of a hash table (unordered_map) gave roughly the same performance.
2017-01-16 15:31:49 -06:00
162638c93c
Merge pull request #1 from QMCPACK/develop
...
Catch up test
2017-01-16 12:18:02 -05:00
a555dc7dbe
Merge pull request #61 from prckent/updating_oxygen_test_scripts
...
Tweak build order
2017-01-16 12:14:02 -05:00
4eef1d1a8d
Tweak build order
2017-01-16 12:12:10 -05:00
924fde62ea
Merge pull request #60 from prckent/updating_oxygen_test_scripts
...
Updated oxygen scripts for ctest/cdash using QMCPACK GitHub repository
2017-01-15 22:01:55 -06:00
e8c851bd54
Updated oxygen scripts for ctest/cdash using QMCPACK GitHub repository
2017-01-15 22:18:00 -05:00
81f7038755
Add single comment to DummyQMC.h for PR test
2017-01-13 12:54:32 -06:00
8bca5b4278
Merge pull request #56 from QMCPACK/a-small-change-in-manual
...
update a line in the manual for github test.
2017-01-12 17:00:17 -05:00
54ee893acb
update a line in the manual for github test.
2017-01-12 15:40:46 -06:00
e78b472074
Merge pull request #55 from prckent/test_docs_update
...
Mention release versions available via git master (small test commit)
2017-01-12 15:32:28 -05:00
a2530c71af
Mention release versions available via git master (small test commit)
2017-01-12 15:28:54 -05:00
e29c4a34b0
Merge pull request #54 from grahamlopez/readme_formatting
...
formatting cleanup for README display on GitHub
2017-01-12 13:31:52 -06:00
5048e7de68
formatting cleanup for README display on GitHub
2017-01-12 14:18:23 -05:00
7bd8bfddbe
Merge pull request #53 from markdewing/manual_git_update
...
Update manual to refer to Github rather than SVN.
2017-01-12 14:00:57 -05:00
Ye Luo