1
0
mirror of https://github.com/paboyle/Grid.git synced 2025-04-25 21:25:56 +01:00

8068 Commits

Author SHA1 Message Date
c4d9aa1a21 Config command that makes GPT happier 2025-02-27 20:12:49 +00:00
6ae809ed40 Print not liked on GPT compile 2025-02-27 20:12:49 +00:00
Peter Boyle
311e2aab3f
Update Accelerator.h 2025-02-26 11:42:52 -05:00
438dfbdb83 Only throw if there is a pending list entry in CommsComplete 2025-02-25 16:57:27 +00:00
b2ce760cf4 Verbose issue with GPT 2025-02-25 16:55:23 +00:00
Muhammad Asif
b1ba209696
Latest upstream with np-su3 patch and modified Sp_WilsonFunfFermionGauge test to be small (#22)
Co-authored-by: Mashy Green <mashy@me.com>

merging no-su3 patch
2025-02-24 11:38:42 +00:00
Muhammad Asif
cb3e529b1e
Merge branch 'paboyle:develop' into develop 2025-02-24 11:29:09 +00:00
Mashy Green
717f647418 added the WilsonFlow patch from upstream PR #471 2025-02-24 08:41:31 +00:00
Mashy Green
98e7418187 Merge remote-tracking branch 'upstream/develop' into gauge_action_deriv 2025-02-24 08:33:05 +00:00
Mashy Green
fe05bf48b1
Improvements to WilsonGaugeAction deriv function (#16)
* patched version + modifications to deriv -> staple in qcd/gauge

* Cleaning up and aligning variable naming between action deriv versions

* Removing the regresion test files that were also in this branch for a clean PR

* Reverting whitespace changes

* Fixing after revering too much!

---------

Co-authored-by: Mashy Green <mashy@me.com>
2025-02-17 18:52:04 +00:00
Mashy Green
d2dd8f54e2 Fixing after revering too much! 2025-02-17 17:32:27 +00:00
Mashy Green
7726ee4b16 Reverting whitespace changes 2025-02-17 17:16:28 +00:00
ba9bbe0221 Bounce MPI through host 2025-02-12 19:34:59 +00:00
4c3dd82d84 CSHIFT with bounce throuhgh Host memory on MPI packets 2025-02-12 19:09:53 +00:00
44e911b5b7 Comment change 2025-02-12 17:37:55 +00:00
a7a16df9d0 GET not put has kinder barrier sequence for NVLINK type access as when
GET is done, I can use it without barrier. Moves a barrier to a nicer
place, overlapped with DtoH DMA
2025-02-12 14:59:28 +00:00
382e0abefd Was issueing a double fence -- the gather also fences 2025-02-12 14:57:28 +00:00
6fdefe5b90 Barrier sequencing if doing "GET" not "PUT" is different.
This is somewhat better timing for Barriers
2025-02-12 14:55:20 +00:00
4788dd8e2e More states in packet progression for GPU non aware MPI 2025-02-12 14:53:57 +00:00
1cc5f221f3 GET not put ordering is better as I know when I've got all MY data 2025-02-12 14:53:05 +00:00
93251bfba0 GET not put for better ordering in the downstream dependent kernels -- I
know when I'm done, so we can move a barrier / handshake between ranks
intranode to a point off critical path
2025-02-12 14:50:21 +00:00
18b79508b8 New line better for pretty print 2025-02-12 14:49:48 +00:00
4de5ed1613 Remove vector view. The std::vector will not inform Memory manager of
deletion and so a stale entry could be left. It is not and should not be
used.
2025-02-12 14:48:46 +00:00
0baaddbe98 Pipeline mode commit on Aurora. 5+ TF/s on 16^3x32 per tile at 384
nodes.
More concurrency/fine grained scheduling is possible.
2025-02-04 19:27:26 +00:00
Mashy Green
355ec76257
Merge pull request #18 from UCL-ARC/bugfix/nvtx
Bugfix/nvtx
2025-02-03 11:05:42 +00:00
b50fb34e71 Perf on Aurora 2025-02-01 18:39:34 +00:00
de84d730ff Fastest run config on Aurora to date 2025-02-01 18:08:40 +00:00
Peter Boyle
c74d11e3d7 PVdagM MG 2025-02-01 11:04:13 -05:00
Christoph Lehner
84cab5e6e7 no comms and log cleanup 2025-02-01 16:37:21 +01:00
c4fc972fec Merge branch 'feature/deprecate-uvm' into develop 2025-01-31 16:32:36 +00:00
8cf809e231 Best results on Aurora so far 2025-01-31 16:14:45 +00:00
94019a922e Significantly better performance on Aurora without using pipeline mode 2025-01-30 16:36:46 +00:00
Mashy Green
4f17c8d081
Merge branch 'paboyle:develop' into bugfix/nvtx 2025-01-29 13:10:12 +00:00
Mashy Green
aaab753982 Reverting to older version of nvtx for Tursa support 2025-01-29 12:57:38 +00:00
d6b2727f86 Pipeline mode getting better -- 2 nodes @ 10TF/s per node on Aurora 2025-01-29 09:22:21 +00:00
74a4f43946 Optional host buffer bounce for no CUDA aware MPI 2025-01-28 15:22:46 +00:00
1caf8b0f86 Rename 2025-01-28 15:22:37 +00:00
Chulwoo Jung
570b72a47b Bugfix. Sorry! 2025-01-21 15:37:39 -05:00
Chulwoo Jung
a5798a89ed Merge branch 'develop' into specflow 2025-01-21 12:13:24 -05:00
Peter Boyle
3f3661a86f Heading towards PVdagM multigrid 2025-01-17 14:33:35 +00:00
Chulwoo Jung
f7e2f9a401 Checking in spectral flow and DWF/Mobius kernel eigenvalue measurement 2025-01-16 20:47:33 +00:00
Chulwoo Jung
2848a9b558 DWF Kernel lanczos working(?) 2025-01-16 01:29:56 +00:00
Mashy Green
d4868991af Fixed wrong lib for NVTX in configure.ac and updated to nvtx3 2025-01-10 14:53:19 +00:00
Mashy Green
e99d42404e Removing the regresion test files that were also in this branch for a clean PR 2024-12-16 16:31:22 +00:00
Mashy Green
3ba019c747 Cleaning up and aligning variable naming between action deriv versions 2024-12-03 15:23:00 +00:00
Mashy Green
47429218bb patched version + modifications to deriv -> staple in qcd/gauge 2024-11-27 16:29:22 +00:00
8fe429346f Dslash testing for reproduce 2024-11-11 23:11:11 +00:00
Peter Boyle
5a4f9bf2e3 Force the ROCM version 2024-10-29 18:12:31 -04:00
Peter Boyle
b91fc1b6b4 Merge branch 'feature/boosted' into feature/deprecate-uvm
Fixed boosted free field test
2024-10-28 16:53:09 -04:00
Peter Boyle
eafc150034 Test fft asserts 2024-10-23 16:46:26 -04:00