Commit Graph

12104 Commits

Author SHA1 Message Date
Ruipeng Li
04037377ea debug for HIP 2022-02-08 14:48:58 -06:00
Ruipeng Li
8ba048c0b5
Forced regeneration of softlinks in shared library builds. (#574)
This PR (copied from #573) added -f to softlink generation to all the makefiles.
2022-02-07 16:54:44 -08:00
Ruipeng Li
e40f8219a3 fix for last merged PR 2022-02-07 18:26:53 -06:00
Quan Bui
734a10fcb7
Mgr setup gpu (#400)
Enable GPU setup for MGR solver.
* Added device specific functionality for interpolation
* Made device and host calls to interpolation consistent
* Edited IJ driver to use GPU capable options for MGR
* Updated saved files for new GPU options
* Updated CMakeLists to support new MGR capabilities

Co-authored-by: Ruipeng Li <li50@llnl.gov>
Co-authored-by: Daniel Osei-Kuffuor <oseikuffuor1@llnl.gov>
2022-02-07 15:54:52 -08:00
Ruipeng Li
790e8e7826
fix cuda 11 build (#569)
This PR fixes CUDA 11 build after merging #549, also adds regression tests (build only) with CUDA 11.
2022-02-02 08:40:15 -08:00
Wayne Mitchell
a7bb784a45
SYCL support for AMG solve phase (#549)
This adds matvec, matrix transpose, and vector operations (axpy, inner product, etc.)
with sycl backend (via oneMKL and oneDPL) for running on Intel GPUs. Thus, the AMG
solve phase can now execute entirely on Intel GPUs.
2022-01-31 16:15:30 -08:00
Victor A. Paludetto Magri
b159c7dd58
Fortran interfaces (#566)
This PR adds Fortran interfaces for hypre_MGR and hypre_ILU. Additionally:

* Add ArrayArray types in `fortran.h`
* Add MGR and ILU options to the fortran interfaces for Krylov solvers
2022-01-31 15:32:14 -08:00
Ruipeng Li
24165eba5a remove hash type from ij.c 2022-01-30 23:49:00 -08:00
Ruipeng Li
8cde1ac442 remove some shared memory in symbl.h 2022-01-30 23:42:03 -08:00
Ruipeng Li
2fdccfaa60 changed default group settings 2022-01-30 17:32:57 -08:00
Ruipeng Li
a43daab879 add symb1/numer1; new binned symbl 2022-01-29 22:40:22 -08:00
Ruipeng Li
ebe3d2bdf3 renamed a marco 2022-01-28 15:50:39 -08:00
Ruipeng Li
38ebda17b3 minor change in a file 2022-01-28 16:22:26 -06:00
Ruipeng Li
40258a8f6b add a new file 2022-01-28 16:09:19 -06:00
Ruipeng Li
6f597695af fix for hip 2022-01-27 22:26:04 -06:00
li50@llnl.gov
9e115e94f5 Merge branch 'deviceomp' of github.com:hypre-space/hypre into spgemm 2022-01-25 22:37:26 -08:00
Ruipeng Li
c8abcff853 fixed compile without stream; enable stream with hip and sycl 2022-01-25 22:17:25 -08:00
li50@llnl.gov
a0f95095f7 Merge branch 'master' of github.com:hypre-space/hypre into deviceomp 2022-01-25 21:44:46 -08:00
Ruipeng Li
f2763da988 change the use of HYPRE_WARP_SIZE in spgemm 2022-01-25 17:29:53 -08:00
Ruipeng Li
543a1e552d remove hash type from runtime 2022-01-25 14:16:43 -08:00
Denis Barbier
dcbee14539
Add convenient CMake alias (#563)
Usually consumers of HYPRE call find_package(HYPRE) and depend on
HYPRE::HYPRE target.  But they could also want to use HYPRE via
add_directory (for instance via a git submodule)) or FetchContent,
in which case they have to depend on HYPRE target.

This alias makes this usage more consistent, all users could then
depend on HYPRE::HYPRE.  See for instance
  https://cmake.org/pipermail/cmake/2018-November/068629.html
2022-01-25 12:06:47 -08:00
Ruipeng Li
ffa79d350f add device shmem properties in device_data 2022-01-25 09:40:15 -08:00
Ruipeng Li
13ba004e75 minor change 2022-01-24 16:53:44 -08:00
Ruipeng Li
8d155b1fd9 binned alg3 2022-01-24 16:46:15 -08:00
Ruipeng Li
3ddf2dd836 minor change 2022-01-24 11:58:58 -08:00
Ruipeng Li
89e037fdac code reorg 2022-01-24 11:55:37 -08:00
Ruipeng Li
d8783345be removed alg 2 in the paper 2022-01-24 09:12:23 -08:00
Ruipeng Li
10e576bff0 completed binned algs. 2022-01-20 21:42:36 -08:00
Ruipeng Li
84fa589671
Redwood sh update (#561)
This PR adds a minor update in runtest.sh for redwood
2022-01-19 21:28:46 -08:00
Ruipeng Li
4c3ef2a0b4
Fortran gpu (#470)
This PR adds GPU examples for FORTRAN users, examples ex5f.f and ex12f.f.
2022-01-19 21:24:05 -08:00
Ruipeng Li
ce54070d76
fixed .saved (#560)
This PR updates files smoother.saved.lassen/ray that were not done in #534.
2022-01-19 16:30:15 -08:00
Ruipeng Li
6bd12e6445 optimization 2022-01-14 14:20:21 -08:00
Ruipeng Li
86710f73dc get number of blocks for bins 2022-01-13 09:01:50 -08:00
Ruipeng Li
514c72be69
add reading x0 from parcsr file back (#548)
This PR adds build_x0_type == 7 (read from parcsr file) back.
2022-01-12 08:57:14 -08:00
Ruipeng Li
bcccb117ef
ldg only for sm >= 35 (#516)
This PR fixes compile issues with CUDA sm_30. See #511
2022-01-12 08:55:58 -08:00
Ruipeng Li
436e09cba2
Early break in CG Eig (#534)
This PR adds early break in CG for eigenvalue estimations.
2022-01-12 08:53:42 -08:00
Wayne Mitchell
c0c265cb20 oneMKL sparse matmat implemented and tested 2022-01-11 21:06:09 +00:00
Wayne Mitchell
2b3704f4c6 Fix cmake to avoid hang on certain oneDPL calls 2022-01-06 01:55:35 +00:00
Wayne Mitchell
a2f3a182ed Merge branch 'sycl' of github.com:hypre-space/hypre into sycl 2022-01-05 23:53:07 +00:00
Wayne Mitchell
fdb59d2a87 Clean up onedpl headers 2022-01-05 23:40:03 +00:00
Wayne Mitchell
852f016e8a Clean up autoconf configure and work on cmake to work with mkl 2022-01-04 22:23:27 +00:00
Wayne Mitchell
9b0207f901 Add dpl lower bound and enable sycl amgdd 2021-12-30 19:23:09 +00:00
Wayne Mitchell
747efa47a6 Missed an error in the merge 2021-12-30 01:47:43 +00:00
Wayne Mitchell
ab14394811 Merge branch 'master' into sycl 2021-12-30 01:19:40 +00:00
Wayne Mitchell
9d1f9c6be7 Cleanup 2021-12-30 00:54:03 +00:00
Wayne Mitchell
69a293d643 Fix oneDPL and matrix transpose
Had to change some compile and linking options to get both MKL
and DPL both working correctly, but this should all work now.
Matrix transpose is now happening on the device for sycl.
2021-12-30 00:41:47 +00:00
Wayne Mitchell
ebdd34d3f9 Some clean up and additional fixes. Pretty sure I am passing all ij regression tests! 2021-12-23 21:22:06 +00:00
Ruipeng Li
87c5e67014 Merge branch 'spgemm' of github.com:hypre-space/hypre into spgemm 2021-12-20 23:08:50 -08:00
Ruipeng Li
cfc7f7bd93
Spgemm bin (#547)
binned SpGEMM algs
2021-12-20 22:49:21 -08:00
Wayne Mitchell
1b9f07192e Add wait commands. Fix some interpolations.
Still a work in progress with some current debugging of
nongalerkin etc. happening, but I wanted to save this here.
2021-12-21 01:26:59 +00:00