be08506 | Balint Joo | 16 October 2017, 16:44:28 UTC | Made cern_gauge_init create dummy file and record xml files, so that QIO conversion may work better | 16 October 2017, 16:44:28 UTC |
384299c | Balint Joo | 29 September 2017, 20:08:20 UTC | Merge of upstream master changes onto devel | 29 September 2017, 20:08:20 UTC |
6bf5d7b | Frank Winter | 23 August 2017, 03:26:21 UTC | Adapt NVVM clover (actually NVPTX) to new qdp-jit/llvm-nvptx interface. GCC specific for now. | 23 August 2017, 14:54:47 UTC |
f2fb4ad | Frank Winter | 23 August 2017, 03:26:29 UTC | Merge branch 'master' of github.com:JeffersonLab/chroma | 23 August 2017, 03:26:29 UTC |
47e6a72 | Frank Winter | 23 August 2017, 03:26:21 UTC | Adapt NVVM clover (actually NVPTX) to new qdp-jit/llvm-nvptx interface. GCC specific for now. | 23 August 2017, 03:26:21 UTC |
716b6b9 | Robert Edwards | 14 August 2017, 22:02:00 UTC | Put in a hard check on the resid. | 14 August 2017, 22:02:00 UTC |
f5985fa | Robert Edwards | 02 August 2017, 14:00:17 UTC | Merge pull request #28 from evanberkowitz/devel Fix annoying chroma-config typo | 02 August 2017, 14:00:17 UTC |
c166492 | Evan Berkowitz | 02 August 2017, 12:49:43 UTC | Fix annoying chroma-config typo | 02 August 2017, 12:49:43 UTC |
e2445e6 | Balint Joo | 18 July 2017, 18:17:19 UTC | Another missing std | 18 July 2017, 18:17:19 UTC |
4a68789 | Balint Joo | 18 July 2017, 18:13:36 UTC | Another missing std | 18 July 2017, 18:13:36 UTC |
1e50feb | Balint Joo | 18 July 2017, 18:13:06 UTC | Another missing std | 18 July 2017, 18:13:06 UTC |
0618847 | Balint Joo | 15 June 2017, 19:16:34 UTC | Added file from previous commit to Makefile.am so it gets installed | 15 June 2017, 19:16:34 UTC |
628ef59 | Balint Joo | 15 June 2017, 19:15:00 UTC | Added a MACRO GetMemoryPtr to wrap the different ways QDPJIT gets memory pointers from cache in QDPJIT/PTX and QDPJIT/LLVMNVVM | 15 June 2017, 19:15:00 UTC |
aee9f4b | Balint Joo | 13 June 2017, 16:36:45 UTC | Added -lqphix_codegen to linking with new QPhiX | 13 June 2017, 16:36:45 UTC |
3ffc72a | Balint Joo | 13 June 2017, 15:43:11 UTC | Merge branch 'devel' of github.com:jeffersonlab/chroma into devel | 13 June 2017, 15:43:11 UTC |
6a2450a | Robert Edwards | 18 May 2017, 12:41:53 UTC | Changed rossi_basis_v2 to be tex-able (not latex). | 18 May 2017, 12:41:53 UTC |
55518ea | Robert Edwards | 18 May 2017, 12:18:45 UTC | Merge branch 'master' of github.com:JeffersonLab/chroma | 18 May 2017, 12:18:45 UTC |
579aa40 | Robert Edwards | 18 May 2017, 12:18:26 UTC | Local mods. | 18 May 2017, 12:18:26 UTC |
150957f | Kostas Orginos | 12 May 2017, 03:24:50 UTC | fix again | 12 May 2017, 03:24:50 UTC |
9e440a8 | Kostas Orginos | 12 May 2017, 00:06:29 UTC | fixed it | 12 May 2017, 00:06:29 UTC |
cd3c7ee | Balint Joo | 11 May 2017, 20:41:38 UTC | Updated News | 11 May 2017, 20:41:38 UTC |
6b798a9 | Balint Joo | 11 May 2017, 19:33:22 UTC | Fixed a typo in configure.ac | 11 May 2017, 19:33:22 UTC |
6dcbb75 | Balint Joo | 11 May 2017, 19:17:13 UTC | Fixed merge conflict | 11 May 2017, 19:17:13 UTC |
89e2a73 | Balint Joo | 11 May 2017, 18:12:06 UTC | Merge branch 'feature/hmc_mg' into devel | 11 May 2017, 18:12:06 UTC |
c4ff98a | Balint Joo | 11 May 2017, 17:28:54 UTC | Merge branch 'devel' of github.com:jeffersonlab/chroma into devel Conflicts: other_libs/cpp_wilson_dslash | 11 May 2017, 17:28:54 UTC |
19d1a2f | Balint Joo | 11 May 2017, 17:09:47 UTC | Either added qdp.h headers or removed math.h includes so that confused isnan/isinf in intel16 does not cause issues/ | 11 May 2017, 17:09:47 UTC |
10ce77b | Kostas Orginos | 10 May 2017, 18:12:01 UTC | ... | 10 May 2017, 18:12:01 UTC |
e6d6695 | Kostas Orginos | 10 May 2017, 18:11:40 UTC | ... | 10 May 2017, 18:11:40 UTC |
d84ea4e | Kostas Orginos | 10 May 2017, 17:40:00 UTC | Merge branch 'master' of github.com:JeffersonLab/chroma | 10 May 2017, 17:40:00 UTC |
a3dfa2b | Kostas Orginos | 10 May 2017, 17:39:47 UTC | added new stouting | 10 May 2017, 17:39:47 UTC |
3dc6927 | Balint Joo | 08 May 2017, 14:52:06 UTC | Removed QPhiX Tune() | 08 May 2017, 14:52:06 UTC |
3e7d046 | Balint Joo | 08 May 2017, 14:46:48 UTC | Merge pull request #15 from martin-ueding/devel Fix copy-and-paste errors | 08 May 2017, 14:46:48 UTC |
443db5d | Kostas Orginos | 23 April 2017, 23:37:47 UTC | added LatticeFermion to nabla | 23 April 2017, 23:37:47 UTC |
e16a2b0 | Kostas Orginos | 23 April 2017, 23:37:28 UTC | added LatticeFermion to nabla | 23 April 2017, 23:37:28 UTC |
cbe1268 | Kostas Orginos | 23 April 2017, 21:58:18 UTC | added LatticePropagator to nabla | 23 April 2017, 21:58:18 UTC |
1c191be | Kostas Orginos | 23 April 2017, 21:57:43 UTC | added LatticePropagator to nabla | 23 April 2017, 21:57:43 UTC |
707f018 | Robert Edwards | 31 March 2017, 17:45:58 UTC | Merge pull request #23 from bglaessle/patch-jlab patch segfaults in readMILC + 2 missing START_CODE() | 31 March 2017, 17:45:58 UTC |
fd42811 | Balint Joo | 31 March 2017, 13:30:53 UTC | Added some basic timers | 31 March 2017, 13:30:53 UTC |
b13270b | Benjamin Glaessle | 31 March 2017, 12:50:07 UTC | fixed typo and 2 missing START_CODE() | 31 March 2017, 12:50:07 UTC |
434e014 | Benjamin Glaessle | 18 July 2016, 16:07:49 UTC | fixed 2nd readMILC double prec bug | 31 March 2017, 12:47:32 UTC |
e279e8f | Benjamin Glaessle | 15 July 2016, 18:05:10 UTC | fixed double prec readMILC segfault (from 2009) | 31 March 2017, 12:46:52 UTC |
3daaf00 | Balint Joo | 27 March 2017, 19:07:43 UTC | Hooray! First working integration of mg_proto Solver. Works on a single node at least including smearing and anisotropy. -- It may or may not give an algorithmic advantage but the solutions are good. | 27 March 2017, 19:07:43 UTC |
5a47094 | Robert Edwards | 27 March 2017, 00:47:06 UTC | Added more fixed precision versions of time-slice writer. | 27 March 2017, 00:47:06 UTC |
172d2fa | Robert Edwards | 27 March 2017, 00:34:41 UTC | Added fixed precision versions. | 27 March 2017, 00:34:41 UTC |
cda379a | Balint Joo | 23 March 2017, 20:39:47 UTC | Initial integration of mg_proto solver. Added: SyssolverLinopShell Paramter files Autoconf/Automake --with-mg-proto= plumbing Test Program t_mgproto.cc Nothing actually being called from mg_proto yet. Next stage is to test XML file reader and actually create the MG solver from MG Proto. | 23 March 2017, 20:39:47 UTC |
7f8075f | Balint Joo | 02 March 2017, 20:20:47 UTC | Merging Robert's changes from master Merge remote-tracking branch 'origin/master' into devel | 02 March 2017, 20:20:47 UTC |
3c828b1 | Robert Edwards | 01 March 2017, 17:56:09 UTC | Test input combines prop and peram calcs. | 01 March 2017, 17:56:09 UTC |
2b9a8d6 | Robert Edwards | 27 February 2017, 05:51:18 UTC | Changed prop_dist_stoch to use a list of sources. | 27 February 2017, 05:51:18 UTC |
0aca5fe | Robert Edwards | 27 February 2017, 04:02:56 UTC | Adding tests to prop_dist_stoch. | 27 February 2017, 04:02:56 UTC |
28aa2e7 | Robert Edwards | 21 February 2017, 15:35:25 UTC | Reverted a change of readcern.cc back to original version. It'll now compile in double but will not compile for Nc != 3. | 21 February 2017, 15:35:25 UTC |
5b183e8 | Robert Edwards | 20 February 2017, 18:39:40 UTC | Added a simple extension. | 20 February 2017, 18:39:40 UTC |
c0e6e59 | Robert Edwards | 20 February 2017, 18:38:42 UTC | Switched to new input form requiring an array of colorvec sources. | 20 February 2017, 18:38:42 UTC |
fc866a6 | Robert Edwards | 20 February 2017, 18:38:02 UTC | Put in a run-time check for Nc in readcern. | 20 February 2017, 18:38:02 UTC |
cb188aa | Robert Edwards | 19 February 2017, 22:35:42 UTC | Also produce a noise source. | 19 February 2017, 22:35:42 UTC |
d910dc4 | Martin Ueding | 14 January 2017, 18:41:08 UTC | Fix WislonFlowGaugeObservables typo (GH-17) | 14 January 2017, 18:41:08 UTC |
fc5d4bd | Martin Ueding | 14 January 2017, 18:40:26 UTC | Add an exception for HEX smearing (GH-16) | 14 January 2017, 18:40:26 UTC |
f669984 | Kostas Orginos | 08 January 2017, 22:45:11 UTC | fixed bug in script | 08 January 2017, 22:45:11 UTC |
2d8a63b | Martin Ueding | 08 January 2017, 11:47:12 UTC | Add a FIXME comment | 08 January 2017, 11:47:12 UTC |
77e597f | Martin Ueding | 25 December 2016, 20:46:31 UTC | Merge branch 'devel' of https://github.com/JeffersonLab/chroma into devel | 25 December 2016, 20:46:31 UTC |
a80a258 | Martin Ueding | 25 December 2016, 20:45:43 UTC | Fix copy-and-paste errors | 25 December 2016, 20:45:43 UTC |
5ffbf4a | Balint Joo | 15 December 2016, 15:33:18 UTC | Updated submodule | 15 December 2016, 15:33:18 UTC |
3b61fd7 | Robert Edwards | 01 December 2016, 19:39:59 UTC | Merge pull request #14 from martin-ueding/master Ignore more build files | 01 December 2016, 19:39:59 UTC |
f564489 | Martin Ueding | 01 December 2016, 11:13:18 UTC | Ignore more build files | 01 December 2016, 11:13:18 UTC |
62cf4d3 | Balint Joo | 03 November 2016, 16:13:14 UTC | Added a test input file for MILC Staggered Source Writing. | 03 November 2016, 16:13:14 UTC |
6126c5d | Balint Joo | 03 November 2016, 16:10:55 UTC | - Templated QIOWriteLatFerm so it can io also LatticeStaggeredFermions in QIO_WRITE_NAMED_OBJECT - added FermToCv function in transf.[h,cc] to allow conversion of a LatticeStaggeredFermion to a ColorVector (peekSpin) - Added a function to I/O a time slice of a LatticeStaggeredPropagator to be used as a source for MILC. | 03 November 2016, 16:10:55 UTC |
234d663 | Arjun Singh Gambhir | 18 October 2016, 23:41:06 UTC | Fixed a bug in reading ThresholdCount in quda_multigrid_clover_params.cc. Also added threshold counter functionality into mdagm_quda_multigrid. If either of the two step solves is above the threshold counter, the subspace is destroyed and remade after the solve. | 18 October 2016, 23:41:06 UTC |
02fb79c | Balint Joo | 18 October 2016, 20:52:32 UTC | Added call to updateMultigridQuda, for subspace persistence. Freeing Gauge fields at end of update Needs feature/multigrid branch of QUDA for now. | 18 October 2016, 20:52:32 UTC |
051b286 | Balint Joo | 18 October 2016, 19:24:40 UTC | Forgot makefile.am | 18 October 2016, 19:24:40 UTC |
48afe46 | Balint Joo | 18 October 2016, 18:57:23 UTC | Factored quda create_subspace and destroy_subspace into quda_mg_utils.h Hopefully fixed crazyness with the erase stuff | 18 October 2016, 18:57:23 UTC |
62e51cf | Balint Joo | 18 October 2016, 17:56:13 UTC | Reverted MDAGM solver to properly delete the subspace and fields on destruction pending fix in QUDA. (Arjun may have working version which persists at the expense of leaking just now) Fixed (?) the QUDA MG Subspace Erasing function | 18 October 2016, 17:56:13 UTC |
4149869 | Balint Joo | 18 October 2016, 00:53:39 UTC | Added destroy_subspace function taking subspace ID as parameter | 18 October 2016, 00:53:39 UTC |
9015543 | Balint Joo | 18 October 2016, 00:22:21 UTC | Arjuns most recent mods. Subspace not persisted, but destroyed and recreated, proving that works. Shows no leaks. Persistance comes next | 18 October 2016, 00:22:21 UTC |
1454f23 | Balint Joo | 17 October 2016, 23:11:40 UTC | Added Default Threshold Count | 17 October 2016, 23:11:40 UTC |
c46fd71 | Balint Joo | 17 October 2016, 22:25:55 UTC | - Reverted QUDA Linop Solver. Now appears stable - Arjun working in syssolver_mdagm_clover_quda_multigrid_w.h - Arjun added threshold stuff to the params | 17 October 2016, 22:25:55 UTC |
470de95 | Arjun Singh Gambhir | 17 October 2016, 14:48:26 UTC | Added some changes from Balint. The quda_mg subspace ID is now always required. He also changed my create and delete subspaces into templates that are out of the class. | 17 October 2016, 14:48:26 UTC |
28f421e | Arjun Singh Gambhir | 13 October 2016, 17:19:29 UTC | Fixed a small bug; included lmdagm header in mdagm_clover_quda_multigrid | 13 October 2016, 17:19:29 UTC |
a5856b1 | Balint Joo | 12 October 2016, 21:02:46 UTC | -uncommented solvy part in the solve with Chrono -added some reporting -UNTESTED!!!! BEWARE!!! -Predictor strategy needs more thought. | 12 October 2016, 21:02:46 UTC |
6d7ef93 | Balint Joo | 07 October 2016, 18:34:54 UTC | - Reworked the default parallel IO condition to check whether an I/O Grid is defined. If not defined, parallel IO won't be enabled by default, and enabling it explicitly in the XML may result in undefined behaviour | 07 October 2016, 18:34:54 UTC |
571cdd4 | Balint Joo | 07 October 2016, 17:51:51 UTC | Added parallel_io option to USQCD DD Pairs prop reader/writer Modified I/O settings for USQCD DD Pair Read/Write, QIO Read/Write, SZINQIO Gauge read, and HMC Gauge Write to that the default parallel I/O setting is determined by the number of nodes in the I/O grid. So if no parallel_io tag is found, but the I/O grid is bigger than 1 node, parallel I/O will be attempted. This feature can be disabled with an explicit <parallel_io>false</parallel_io> tagset. | 07 October 2016, 17:51:51 UTC |
e5cfadb | Balint Joo | 06 October 2016, 13:21:08 UTC | Added routine in the gauge transf to apply an existing gauge rotation to a gauge field. - currently only LatticeColorMatrix (ie floating precision type) is supported. If you want DP, run in DP Added measurements to gauge_transf.ini.xml to perform the test, and some plaquette measurements so we can verify correctness Adds feature request from A. W-L. Ready to merge into devel | 06 October 2016, 13:21:08 UTC |
c948a59 | Balint Joo | 05 October 2016, 20:26:14 UTC | Adopted new CloverTermT - style | 05 October 2016, 20:26:14 UTC |
6e10ce4 | Balint Joo | 05 October 2016, 19:46:21 UTC | Fixed conflicts to continue merge | 05 October 2016, 19:46:21 UTC |
0f983b7 | Balint Joo | 26 September 2016, 17:16:00 UTC | The iter refine BICGSTAB solver now works for MDAGM | 26 September 2016, 17:16:00 UTC |
74381b8 | Balint Joo | 21 September 2016, 06:03:26 UTC | Updated for better time reporting in iter refine MDAGM solver | 21 September 2016, 06:03:26 UTC |
ef950ca | Balint Joo | 21 September 2016, 02:57:27 UTC | Fixed not setting n_blas_simt. T_monomial force now worked with iter refine solver. Queued Short Traj Test | 21 September 2016, 02:57:27 UTC |
e10991d | Balint Joo | 20 September 2016, 22:12:28 UTC | Fixed up the iter refine solver so that it compiles | 20 September 2016, 22:12:28 UTC |
8e9fe62 | Balint Joo | 14 September 2016, 18:00:13 UTC | Added -lcuda link flag now needed for QUDA with P2P Fix | 14 September 2016, 18:00:13 UTC |
d2ee517 | Balint Joo | 14 September 2016, 15:10:56 UTC | Updated other_libs_cpp_wilson_dslash to devel branch Started working on inner-outer-MDAGM solver NOT YET COMPLETED | 14 September 2016, 15:10:56 UTC |
e1127a6 | Balint Joo | 13 September 2016, 13:03:25 UTC | Updated to use QDP::Allocator instead of Pool Allocator - Added options --enable-static-packed-gauge -- Save on allocations by making Dslash packed gauge static. Dangerous tho. Default off - --enable-fused-clover-deriv-loops - Added Option to use fused site loop from Jacques Bloch of Regensburg university in deriv_loops NB: His original code modified: Not separately specialized for single/double, but templated Mat mults come from QDP++ for now -- enable-ittnotify - defines CHROMA_USE_ITTNOTIFY which can be used to guard ITTNOTIFY options __itt_pause/_itt_resume User still needs to set additional link flags to link and use VTune | 13 September 2016, 13:03:25 UTC |
8e92ad7 | Balint Joo | 08 September 2016, 13:02:30 UTC | Merge branch 'devel' into tbb_testing | 08 September 2016, 13:02:30 UTC |
c543791 | Balint Joo | 08 September 2016, 13:01:52 UTC | Readded Y vector unpack for non-QDP-JIT to fix chronological solver hang/nan bug | 08 September 2016, 13:01:52 UTC |
d39d435 | Balint Joo | 06 September 2016, 20:22:01 UTC | Updated clover_term base and lwldslash_w_qphix.h for static ness. Removed const qualifier in const int return types as the compiler claimed they were meaningless | 06 September 2016, 20:22:01 UTC |
f7dff6c | Arjun Singh Gambhir | 06 September 2016, 20:15:17 UTC | Need to update aggregate file for new erase measurement. | 06 September 2016, 20:15:17 UTC |
4154461 | Arjun Singh Gambhir | 06 September 2016, 20:11:50 UTC | Added Inline Erase measurement to delete subspace and other quda fields. This hasn't been tested with an XML yet, also code may have to be re-arranged further so quda fields are deleted after every update while the subspace remains with the freq param that's in the measurement code. There are a few other points that need to be cleaned up. QUDA MG params has LoadSubspace which can probably be deleted. Both the linop and mdagm have unused subspace functions such as reset_subspace() and correspondingly reset_subspace(), delete subspace(). Finally, the save or recover subspace logic is currently in the mdagm constructor, this may have to be migrated to the solve. Many things to tidy up but the bulk of the code is finished and working. | 06 September 2016, 20:11:50 UTC |
f525e13 | Arjun Singh Gambhir | 05 September 2016, 20:18:56 UTC | Fixed a bug related to copying XML params. Also finished MG persistence logic. This works with my basic XML and the subspace does indeed persist through the trajectory! There is a hiccup with freeing memory at the end I have to fix. The next step will be an inline task that can delete the subspace from the Named Object Storage and also to add frequency functionaly so that the subspace may be refreshed after some number of counts. | 05 September 2016, 20:18:56 UTC |
1875e33 | Balint Joo | 03 September 2016, 20:40:34 UTC | Added Clover Force Term fused loop from Jacques | 03 September 2016, 20:40:34 UTC |
86c354e | Balint Joo | 03 September 2016, 19:14:15 UTC | Added New Pool allocation to clover_term_qdp_w.h Reworked cpp_wilson_dslash. This branch should use its qdp_allocator branch Reworked t_monomial_force - to make it a simple force testing harness | 03 September 2016, 19:14:15 UTC |
18c441d | Arjun Singh Gambhir | 01 September 2016, 23:08:07 UTC | Added previous subspace persistence to mdagm solver as well. Began implementation of the NamedObject Mapping. Next I have to figure out XML propagation. | 01 September 2016, 23:08:07 UTC |
78a7aeb | Arjun Singh Gambhir | 31 August 2016, 23:33:02 UTC | Moved all MG related parameter setup to a create_subspace() function that returns a void*. Functions also added for delete_subspace() and reset_subspace(). I haven't touched the NamedObjectMap yet, but that will be the next step. | 31 August 2016, 23:33:02 UTC |
a09e89a | Arjun Singh Gambhir | 26 August 2016, 17:25:50 UTC | Updated linop, param, and mdagm files to match newest versoin of quda. Also Baint's XML files which specify smoother applications in arrays were broken with this version of the code, fixed now. | 26 August 2016, 17:25:50 UTC |