portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2024-09-20 09:15:38 +01:00
Code Releases Activity
3,562 Commits 86 Branches 14 Tags 74 MiB
aaf37ee4d7
Commit Graph

80 Commits

Author SHA1 Message Date
paboyle
276f113f28 IO uses master boss node for metadata. 2018-03-30 16:17:05 +01:00
paboyle
bda97212a9 Simplify proliferation of comms layers 2018-01-08 11:29:20 +00:00
paboyle
3729c7a7a6 Clean up of test 2017-12-05 13:07:31 +00:00
paboyle
e6a3e375cf Debug 2017-11-27 15:10:22 +00:00
paboyle
4987edbd44 Merge branch 'develop' of https://github.com/paboyle/Grid into develop 2017-11-27 12:34:56 +00:00
paboyle
ad140bb6e7 Clean on multinode target after split 1 1 2 4 -> 1 1 2 2 2017-11-27 12:34:25 +00:00
azusayamaguchi
e55397bc13 Staggerd cg 2017-11-24 14:18:30 +00:00
paboyle
94b8fb5686 Debug in progress 2017-11-19 01:39:04 +00:00
paboyle
78e8704eac Shaking out 2017-10-30 00:25:31 +00:00
paboyle
67131d82f2 Get subrank info from communicator constructor 2017-10-30 00:24:11 +00:00
paboyle
615a9448b9 Extended sub comm supported 2017-10-30 00:23:34 +00:00
paboyle
00164f5ce5 : 2017-10-30 00:22:52 +00:00
paboyle
5bf42e1e15 Update 2017-10-30 00:05:21 +00:00
paboyle
84b441800f Merge branch 'develop' into feature/lanczos-reorg 2017-10-27 14:21:38 +01:00
Azusa Yamaguchi
034de160bf Staggered updates : Schur fixed and added a unit test for Test_staggered_cg_schur.cc giving stronger check 2017-10-26 20:58:46 +01:00
paboyle
2db05ac214 Test for split/unsplit in isolation 2017-10-26 07:48:03 +01:00
paboyle
d577211cc3 Relax stoppign condition 2017-10-25 23:57:54 +01:00
paboyle
f4336e480a Faster converge time 2017-10-25 23:53:44 +01:00
paboyle
9fe6ac71ea Starting reorg of Blocked lanczos 2017-10-11 10:12:07 +01:00
paboyle
bf58557fb1 Block compressed Lanczos 2017-10-10 14:15:11 +01:00
paboyle
10cb37f504 Merge branch 'develop' of https://github.com/paboyle/Grid into develop 2017-10-10 14:09:44 +01:00
Azusa Yamaguchi
1374c943d4 Correct Schur operator called 2017-10-10 13:59:50 +01:00
paboyle
d1c6288c5f Merge branch 'develop' of https://github.com/paboyle/Grid into develop 2017-10-10 13:38:40 +01:00
Azusa Yamaguchi
bb7378cfc3 Schur for staggered 2017-10-10 12:02:18 +01:00
paboyle
153672d8ec Split CG testing 2017-10-09 23:20:58 +01:00
Azusa Yamaguchi
09f4cdb11e Merge branch 'develop' of https://github.com/paboyle/Grid into feature/staggering 2017-10-04 10:51:16 +01:00
Azusa Yamaguchi
1e54882f71 Stagger 2017-10-04 10:51:06 +01:00
paboyle
e5c19e1fd7 RB constructor change 2017-10-02 12:25:52 +01:00
paboyle
a11d0a33d1 Merge branch 'feature/dwf-multirhs' of https://github.com/paboyle/Grid into feature/dwf-multirhs 2017-10-02 11:42:07 +01:00
paboyle
4f8b6f26b4 Merge branch 'develop' into feature/dwf-multirhs 2017-10-02 11:41:49 +01:00
paboyle
073525c5b3 Small patch from cori 2017-10-02 03:38:21 -07:00
paboyle
5918769f97 Subtle Naik term bug updated in Stencil; less on logical && with a function call on right 2017-09-16 12:51:26 +01:00
Azusa Yamaguchi
d9cd4f0273 Staggered multinode block cg debugged. Missing global sum. 
Code stalls and resumes on KNL at cambridge. Curious.

CG iterations 23ms each, then 3200 ms pauses. Mean bandwidth reports
as 200MB/s. Comms dominant in the report. However, the time behaviour suggests it
is *bursty*.... Could be swap to disk?
 2017-08-23 15:07:18 +01:00
paboyle
ac740f73ce Works on Cori 2017-07-02 16:47:58 -07:00
paboyle
dee68fc728 IO working multiple nodes again. Strategy of all nodes writing metadata is unsafe. 
Only one rank should do this. must identify this rank. Means pass communicator to the
Objects.
 2017-07-02 23:33:48 +01:00
paboyle
57002924bc NERSC shakeout of this 2017-07-02 14:58:30 -07:00
paboyle
600d7ddc2e Proof of concept : Multi RHS solver, running independent solves on different ranks 2017-06-22 18:54:34 +01:00
paboyle
6ebf9f15b7 Splitting communicators first cut 2017-06-22 08:14:34 +01:00
paboyle
1d7aa673a4 Include BlockCG by default 2017-06-21 21:08:53 +01:00
paboyle
7e35286860 Simplified lanczos, added Eigen diagonalisation. 
Curious if we can deprecate dependencly on BLAS.
Will see when we get 48^3 running on our BG/Q port
 2017-06-21 02:26:03 +01:00
paboyle
0486ff8e79 Improved the lancos 2017-06-20 18:46:01 +01:00
Azusa Yamaguchi
e9cc21900f Block solver complete for staggered. Now stable on mass 0.003 and 
gives 8x (!) speed up on Haswell laptop vs. standard CG for 8 RHS solves.

166 iterations vs. 537 iterations so algorithmic gain + 2x in flop rate gain.

Better than a slap in the face with a wet kipper.
 2017-06-20 12:37:41 +01:00
Azusa Yamaguchi
0a8faac271 Fix make tests compile 2017-06-19 22:54:18 +01:00
Azusa Yamaguchi
cfe3cd76d1 Block solver improvements 2017-06-19 14:04:21 +01:00
Guido Cossu
20999c1370 Merge branch 'develop' into feature/hmc_generalise 2017-05-05 12:47:17 +01:00
paboyle
33f0ed1a33 No compile fix 2017-05-05 11:04:30 +01:00
Guido Cossu
3344788fa1 Merge branch 'develop' into feature/hmc_generalise 2017-05-01 12:13:56 +01:00
paboyle
8e161152e4 MultiRHS solver improvements with slice operations moved into lattice and sped up. 
Block solver requires a lot of performance work.
 2017-04-18 10:51:55 +01:00
paboyle
bf516c3b81 higher precision reduction variables in norm and inner product 2017-04-15 12:27:28 +01:00
paboyle
d80d802f9d MultiRHS solver test 2017-04-10 00:12:12 +09:00