mmorale3
39299ecb60
pull upstream and bug fix
2019-03-25 12:28:31 -07:00
mmorale3
83fcf5151b
fixed real build, needs unit testing and real live tests
2019-03-22 14:39:46 -07:00
Ye Luo
07a909898d
Merge branch 'develop' into use-cublas
2019-03-20 20:17:03 -05:00
Ye Luo
00c5156c22
Update CMakeLists.txt.
2019-03-20 17:03:03 -05:00
Peter Doak
9574ad6328
Merge branch 'develop' into offload-spline
2019-03-20 16:44:19 -04:00
Ye Luo
ab76df2870
Merge remote-tracking branch 'origin/develop' into use-cublas
2019-03-18 09:58:01 -05:00
mmorale3
07ee9ba05f
pull upstream and conflict resolution
2019-03-15 15:51:03 -07:00
Ye Luo
511a353ca8
More cleaning.
2019-03-13 18:50:29 -05:00
Ye Luo
13aaeb4715
Rename macro ENABLE_CUBLAS to ENABLE_CUDA.
2019-03-12 18:43:48 -05:00
Ye Luo
24259b7940
A bit clean up.
2019-03-10 11:48:21 -05:00
Ye Luo
76983495b6
Merge branch 'single-det-class-inheritance' into use-cublas
2019-03-09 21:19:58 -06:00
Ye Luo
d7bdfddbf0
Merge remote-tracking branch 'origin/develop' into offload-spline
2019-03-07 08:39:09 -06:00
Ye Luo
84d4cb0d1d
Check compiler and libstdc++ compatibility
2019-03-05 13:52:02 -06:00
mmorale3
204d0ff471
several bug fixes on gpu mixed precision
2019-03-04 10:28:28 -08:00
Ye Luo
ee972ad24f
Rename QMC_CUDA_NEXT to ENABLE_CUBLAS=1
2019-03-01 23:16:43 -06:00
Ye Luo
a480b564b7
Merge branch 'develop' into use-cublas-try-merge
2019-03-01 20:51:11 -06:00
mmorale3
c4a8e018ed
working on mixed precision
2019-03-01 13:35:20 -08:00
Ye Luo
b11347191f
Add ENABLE_OFFLOAD
2019-02-04 21:50:09 -06:00
Ye Luo
0c55da571c
Turn off AFQMC by default.
...
Explicitly turn on AFQMC in CI GNU builds.
2019-02-04 10:49:53 -06:00
Ye Luo
4e2327ebd3
Remove any mention of Xeon Phi accelerator.
2019-02-01 22:00:54 -06:00
Mark Dewing
aef797a16c
Remove QMC_BUILD_LEVEL
...
The default had been set to 3, and was only changed to build the sandbox apps.
Remove the ifdefs in most places, and replace with QMC_BUILD_SANDBOX_ONLY
where appropriate.
Fixes #1051 .
2019-01-29 18:06:37 -06:00
Paul Kent
6afa0ff2af
Remove more legacY
2018-12-27 09:25:52 -05:00
Ye Luo
eac4395894
Do not turn on AFQMC with QMC_CUDA.
2018-12-22 08:03:53 -06:00
Ye Luo
89ead3203a
Enable AFQMC by default with the MPI complex full precision build.
2018-12-22 06:20:24 -06:00
Paul Kent
c6f02a8640
Increase version number to 3.6.0, update release notes
2018-12-18 20:37:00 -05:00
Mark Dewing
ca966720ca
Set minimum boost version to 1.61.0
...
Addresses #1254
This is for AFQMC. The real-space version may have more relaxed version requirements,
but this version seems available most everywhere and for simplicity a single
version requirement is added.
2018-12-14 15:27:51 -06:00
Ye Luo
a63f519a42
Merge remote-tracking branch 'github/develop' into merge_afqmc
2018-12-13 22:04:47 -06:00
Mark Dewing
997aa4124e
Add configuration check for AFQMC and QMC_COMPLEX
...
AFQMC only works with QMC_COMPLEX=1 for now.
2018-12-13 20:53:24 -06:00
Mark Dewing
361d2ef36a
Merge branch 'develop' into nx_release_1_6
2018-12-13 10:48:22 -06:00
Mark Dewing
43c38f8963
Fix PGI compiler build with GCC 7
...
Move the CMake C++ standard settings before the compiler-specific configuration files.
The compiler configs can then override these settings if necessary.
2018-12-11 17:15:22 -06:00
Jaron Krogel
75f56e9645
nexus: add option enable/disable nexus install in cmake
2018-12-11 17:23:02 -05:00
Jaron Krogel
e6494c152c
nexus: defer install and add flags to suppress rc file path update
2018-12-11 13:51:53 -05:00
Jaron Krogel
29b6928feb
nexus: fix cmake install path
2018-12-11 12:43:09 -05:00
Jaron Krogel
4440a80c6b
nexus: add nexus install to cmake
2018-12-11 12:05:40 -05:00
Mark Dewing
a5fef58368
Add install for qmcpack binaries
...
Also install the qmcpack.settings file.
2018-12-04 21:40:54 -06:00
Ye Luo
942b956f26
Add allocator based on cudamalloc.
2018-11-08 11:24:08 -06:00
Ye Luo
d18f71342a
Add CUDAallocator allocator.
2018-11-08 09:07:45 -06:00
Mark Dewing
d149380b7d
Initial support for PGI compiler
2018-11-02 15:56:44 -05:00
Ye Luo
2e908cd8f3
More language correction.
2018-10-31 16:34:21 -05:00
Mark Dewing
2ab0838130
Remove unused code for C++14 check.
2018-10-30 14:12:35 -05:00
Mark Dewing
9afbfa74d6
Test that a C++14 library is configured.
...
Also try to give some advice based on which compiler is detected.
2018-10-25 17:22:07 -05:00
Mark Dewing
4122b264b9
Flip the switch to require C++ 14
2018-10-25 13:27:00 -05:00
Mark Dewing
b3133b60d3
Fix compile on BG/Q (Mira)
...
The required minimum cmake version also sets the policies according to that
cmake version. The CMP0060 policy changed in cmake 3.3.
This policy change caused the libraries returned from finding the HDF library
to fail to link, because CMake previously altered the form on the link line,
and the new behavior does not.
The policy cannot be set in the toolchain file because of warnings related
to policy scope and CMP0011.
2018-10-23 21:31:53 +00:00
Mark Dewing
d29e5af0c7
Increase cmake minimum to 3.6.0.
...
Remove the unneeded cmake version tests.
Switch to the cmake mechanism for specifying the C++11 standard
3.6.0 is the first version that gets the C++11 flags correct for the Intel compiler
Addresses #832
2018-10-19 17:03:50 -05:00
Paul Kent
f5c7747afb
Correct spelling errors across entire source
2018-10-10 19:58:16 -04:00
Ye Luo
0b8d1662f5
Cache ENABLE_SOA in CMakeLists and print in header
2018-10-08 10:54:48 -05:00
Ye Luo
3c40bfd1fb
Remove QMC_SK_USE_RECURSIVE.
2018-09-01 17:15:14 -05:00
Paul Kent
12a520fe84
Changelog for v3.5.0
2018-08-01 15:45:56 -04:00
Peter Doak
4504d64238
will drop the deprecated and inefficient link_libraries another day
2018-07-05 19:45:59 -04:00
Peter Doak
a77ed4f2a6
Unbreaking CXX11 check. Needed appleclang for cuda and its horrible.
...
But that is a different issue.
2018-07-05 19:21:32 -04:00