Commit Graph
100 Commits
Author SHA1 Message Date
Peter Boyle 4efa042f50 C++17 change 2023-10-24 10:57:50 -04:00
Peter Boyle c7cb37e970 c++17 accepted 2023-10-24 10:57:24 -04:00
Peter Boyle d34b207eab Avoid HIP warnings 2023-10-24 10:57:04 -04:00
Peter Boyle 0e6fa6f6b8 DOn't need the Cshift for the period optimisation 2023-10-24 10:56:31 -04:00
Peter Boyle 38b87de53f This works around a stacksize limit on AMD GPU 2023-10-24 10:56:07 -04:00
Peter Boyle aa5047a9e4 Faster blockProject blockPromote 2023-10-24 10:49:55 -04:00
Peter Boyle 24b6ee0df9 M4 file 2023-10-24 10:36:48 -04:00
Peter Boyle 1e79cc9cbe Avoid compiler error 2023-10-24 10:36:09 -04:00
Peter Boyle b3925df9c3 Verbose on CPU-GPU xfer, remove performance by default 2023-10-24 10:25:01 -04:00
Peter Boyle 351795ac3a Better messaging 2023-10-20 19:33:04 -04:00
Peter Boyle 9c9c42d0df Tests on frontier with real speed up . 3.5x on 16^3 at mq=0.01 2023-10-20 19:27:13 -04:00
Peter Boyle b6ad1bafc7 Normal memory SendToRecvFrom asynchronous for use in general stencil
code
2023-10-20 19:27:13 -04:00
Peter Boyle a5ca40f446 Better verbose -- track CPU GPU motion under --log Memory, others go to
debug output stream
2023-10-20 19:27:13 -04:00
Peter Boyle 9ab54c5565 Overlap comms & data copy/buffer assembly in Ghost zone exchange 2023-10-20 19:27:13 -04:00
Peter Boyle 4341d96bde Massively sped up coarse grid mult, comms
Save 3ms spend (60% of time !) on cudaMalloc !!
2023-10-20 19:27:13 -04:00
Peter Boyle 5fac47a26d Faster halo exchange 2023-10-20 19:27:13 -04:00
Peter Boyle e064f17346 Faster halo exchange 2023-10-20 19:27:13 -04:00
Peter Boyle afe10ba2a2 More digits 2023-10-20 19:27:13 -04:00
Peter Boyle 7cc3435ba8 Imporved General coarsened matrix 2023-10-20 19:27:13 -04:00
Peter Boyle 541772313c Verbosity 2023-10-20 19:27:13 -04:00
Peter Boyle 3747494a09 Notify delet public 2023-10-20 19:27:13 -04:00
Peter Boyle f2b98d0dcc Const safety 2023-10-20 19:27:13 -04:00
Peter Boyle 80471bf762 Alternate implementation involving face operations 2023-10-20 19:27:13 -04:00
Peter Boyle a06f63c110 Improved I/O and non-lexico option exposed to SciDAC format 2023-10-20 19:27:13 -04:00
Peter Boyle 0ae4478cd9 Checkpoint the subspace and ldop 2023-10-20 19:27:13 -04:00
Peter Boyle ae4e705e09 Use random vec as easier for debug 2023-10-20 19:27:13 -04:00
Peter Boyle f5dcea9dbf Updates for Frontier 2023-10-20 19:27:12 -04:00
Peter Boyle 2207309f8a Spack rules 2023-10-16 18:38:24 -04:00
Peter Boyle 2111e7ab5f Run at physical mass 2023-10-06 21:20:21 -04:00
Peter Boyle d29abfdcaf Transfer code to Frontier now 2023-10-06 21:03:34 -04:00
Peter Boyle a751c42cc5 Checkpoint restore the setup 2023-10-06 21:03:08 -04:00
Peter Boyle 6a3bc9865e Verbose change 2023-10-06 21:02:04 -04:00
Peter Boyle 4d5f7e4377 Verbose change 2023-10-06 21:01:37 -04:00
Peter Boyle 78b117fb78 Comment fix 2023-10-06 21:01:15 -04:00
Peter Boyle ded63a1319 Verbose change/pretty print 2023-10-06 21:00:53 -04:00
Peter Boyle df3e4d1e9c Return fix 2023-10-06 21:00:21 -04:00
Peter Boyle b58fd80379 I/O for coarse op and reorganise multigrid headers 2023-10-06 13:43:46 -04:00
Peter Boyle 7f6e0f57d0 No IO in file 2023-10-06 13:39:53 -04:00
Peter Boyle cae27678d8 gpermute 2023-10-06 13:39:19 -04:00
Peter Boyle 48ff655bad Slightly less verbose 2023-10-06 10:47:52 -04:00
Peter Boyle 2525ad4623 Slight clean up 2023-10-06 10:47:32 -04:00
Peter Boyle e7020017c5 Reorganise multigrid 2023-10-06 10:47:12 -04:00
Peter Boyle eacebfad74 Reorganise multigrid into multiple headers 2023-10-06 10:46:21 -04:00
Peter Boyle 3bc2da5321 Merge branch 'feature/scidac-wp1' of https://github.com/paboyle/Grid into feature/scidac-wp1 2023-10-05 16:57:59 -04:00
Peter Boyle 2d710d6bfd Optimised parameters for 16^3 2023-10-05 16:56:55 -04:00
Peter Boyle 6532b7f32b Eliminate older inefficient coarsening implementation 2023-10-05 16:56:15 -04:00
Peter Boyle 7b41b92d99 Only need to bad non-local dimensions 2023-10-05 16:55:48 -04:00
Peter Boyle dd557af84b ADEF1 and ADEF2 2 level CG 2023-10-05 16:55:19 -04:00
Peter Boyle 59b9d0e030 coalesceRead the blockSum 2023-10-05 16:54:48 -04:00
Peter Boyle b82eee4733 Hermitian dealing with 2023-10-05 16:54:14 -04:00
Peter Boyle 6a87487544 Running on Frontier, fix RNG big volume y2k, affecting 5D RNG 2023-10-05 16:50:59 -04:00
Peter Boyle fcf5023845 Running on Frontier 2023-10-05 16:50:59 -04:00
Peter Boyle c8adad6d8b First runs on Summit. PopulateAdag needs work 2023-10-05 16:50:54 -04:00
Peter Boyle 737d3ffb98 ADEF1 and 1 hop projection 2023-10-03 14:22:18 -04:00
Peter Boyle b01e67bab1 coalescedReadGeneralPermute now working 2023-10-02 17:46:57 -04:00
Peter Boyle 8a70314f54 Merge branch 'develop' into feature/scidac-wp1 2023-10-02 17:24:55 -04:00
Peter Boyle afc316f501 Rename headers 2023-10-02 16:25:11 -04:00
Peter Boyle f14bfd5c1b Relocate sub includes 2023-10-02 16:23:38 -04:00
Peter Boyle c5f1420dea Merge remote-tracking branch 'LupoA/develop' into LupoA-develop 2023-10-02 16:22:35 -04:00
Peter Boyle 36ae6e5aba Fastest GPU version.
Need to work on the PaddedCell now to make much faster
2023-09-29 18:26:51 -04:00
Peter Boyle 9db585cfeb Temporary commit while optimisation is carried out 2023-09-29 17:11:35 -04:00
Peter Boyle c564611ba7 Annoying hack that is useful to preserve for profiling 2023-09-29 17:11:12 -04:00
Peter Boyle e187bcb85c Updating 2023-09-29 17:10:17 -04:00
Peter Boyle be18ffe3b4 Further tuning and lanczos 2023-09-27 16:21:58 -04:00
Peter Boyle 0d63dce4e2 Timing info 2023-09-27 16:21:14 -04:00
Peter Boyle 26b30e1551 Flop count and projection to nearest neighbour (keeps redundant flops) 2023-09-27 16:20:11 -04:00
Peter Boyle 7fc58ac293 Verbose subspace init 2023-09-27 16:19:45 -04:00
Peter Boyle 3a86cce8c1 Compile 2023-09-27 16:19:18 -04:00
Peter Boyle 80359e0d49 Bland SYCL compile 2023-09-26 13:20:27 -07:00
Peter Boyle 3d437c5cc4 Making SYCL happy 2023-09-26 13:19:42 -07:00
Peter Boyle 37884d369f Coarse space is expensive, but gives a speed up in fine matrix multiplies now.
Down to optimisation
2023-09-25 17:24:19 -04:00
Peter Boyle 9246e653cd Basic non-local coarsening of operator test 2023-09-25 17:20:58 -04:00
Peter Boyle 64283c8673 Normal equations becomes linear function for easy base class pass aroudn 2023-09-25 17:19:39 -04:00
Peter Boyle 755002da9c Comparison convenience 2023-09-25 17:16:33 -04:00
Peter Boyle 31b8e8b437 Better messaging 2023-09-25 17:16:14 -04:00
Peter Boyle 0ec0de97e6 Adef2 implemented and working in an HDCG like context 2023-09-25 17:15:03 -04:00
Peter Boyle 6c3ade5d89 Improved the coarsening 2023-09-25 17:14:40 -04:00
Peter Boyle 980c5f9a34 Update chebyshev setup 2023-09-25 17:12:22 -04:00
Peter Boyle 471ca5f281 Power method more iterations 2023-09-07 10:55:05 -04:00
Peter Boyle e82ddcff5d Working getting closer to HDCG but some low level engineering work still needed
+ MUCH work on optimisation
2023-09-07 10:53:51 -04:00
Peter Boyle b9dcad89e8 Test cases for coarsening with non-local stencil 2023-09-07 10:53:22 -04:00
Peter Boyle 993f43ef4a Even odd use case 2023-09-07 10:53:06 -04:00
Peter Boyle 2b43308208 First cut non-local coarsening 2023-08-25 17:38:07 -04:00
Peter Boyle 04a1ac3a76 First cut for non-local coarsening 2023-08-25 17:37:38 -04:00
Peter Boyle 990b8798bd Merge remote-tracking branch 'refs/remotes/origin/develop' into develop 2023-08-25 17:36:45 -04:00
Peter Boyle b334a73a44 Stencil improvement 2023-08-25 17:35:10 -04:00
Peter Boyle 5d113d1c70 Odd address sanitizer complain 2023-08-25 17:34:18 -04:00
Peter Boyle c14977aeab Random vector option for test purposes 2023-08-25 17:33:31 -04:00
Peter Boyle 3e94838204 Spread out improvement 2023-08-25 17:31:28 -04:00
Peter Boyle c0a0b8ca62 NEON and address sanitiser 2023-08-25 17:30:30 -04:00
Peter Boyle b8a7004365 Partial fraction test 2023-08-14 15:17:03 -04:00
Peter Boyle 6b150961fe Better script 2023-06-23 18:09:25 +03:00
Peter Boyle 5bafcaedfa Merge branch 'develop' of https://github.com/paboyle/Grid into develop 2023-06-22 19:59:45 +03:00
Peter Boyle bfeceae708 FTHMC 2023-06-22 12:58:18 -04:00
Peter Boyle eacb66591f Config command 2023-06-22 19:56:40 +03:00
Peter Boyle fadaa85626 Update 2023-06-22 19:56:27 +03:00
Peter Boyle 02a5b0d786 Updating run during testing 2023-06-22 19:52:46 +03:00
Peter Boyle 0e2141442a Dennis says broken 2023-06-22 19:19:51 +03:00
Peter Boyle 769eb0eecb Precision coverage 2023-06-22 19:19:20 +03:00
Peter Boyle d72e914cf0 Profiling temporary code until optimised 2023-06-15 10:43:04 -04:00