1
0
mirror of https://github.com/paboyle/Grid.git synced 2025-06-24 10:42:03 +01:00

Commit Graph

  • 124bf4d829 git ref in config summary Antonin Portelli 2017-05-02 19:41:01 +01:00
  • e8e56b3414 Config summary saved in git-config Antonin Portelli 2017-05-02 19:40:47 +01:00
  • 89c430136d grid-config program Antonin Portelli 2017-05-02 19:13:13 +01:00
  • ea9aef7baa New header for standard headers (was an issue with Remez.h and external compilation) Antonin Portelli 2017-05-02 18:26:11 +01:00
  • c9e9e8061d Merge branch 'feature/hadrons' into develop Antonin Portelli 2017-05-02 18:23:47 +01:00
  • 453cf2a1c6 Moving the topological charge outside the HMC related routines Guido Cossu 2017-05-02 14:40:12 +01:00
  • de7bbfa5f9 Adding ParameterFile option for the HMC Guido Cossu 2017-05-02 12:16:16 +01:00
  • 867fe93018 First Rotate reorg done. Chulwoo Jung 2017-05-02 01:26:22 -04:00
  • 09651c3326 Checking in before rearranging Lanczos Chulwoo Jung 2017-05-02 00:47:18 -04:00
  • dda8d77c87 Merge branch 'feature/hadrons' into feature/rare_kaon Antonin Portelli 2017-05-01 17:50:57 +01:00
  • aa29f4346a Hadrons: weird bus error with recent macOS clang Antonin Portelli 2017-05-01 17:49:08 +01:00
  • f87f2a3f8b Merge branch 'develop' of https://github.com/paboyle/Grid into feature/Lanczos Chulwoo Jung 2017-05-01 12:00:47 -04:00
  • 86116dbed6 Adding boundary condition switch (compile time) for the Mobius HMC example Guido Cossu 2017-05-01 16:33:11 +01:00
  • 7bd31e3f7c Adding external file support in the Mobius example (JSON) Guido Cossu 2017-05-01 16:30:24 +01:00
  • 74f451715f Fix for Mac compilation on the size_t uint64_t types Guido Cossu 2017-05-01 15:12:07 +01:00
  • 655be8ed76 Adding tests for the mobius operator Guido Cossu 2017-05-01 14:42:16 +01:00
  • 4063238943 Adding HMC test file example for Mobius + smearing Guido Cossu 2017-05-01 13:44:00 +01:00
  • 3344788fa1 Merge branch 'develop' into feature/hmc_generalise Guido Cossu 2017-05-01 12:13:56 +01:00
  • 62a64d9108 EO support, wip Guido Cossu 2017-05-01 11:06:21 +01:00
  • 49331a3e72 Minor improvements to Ward Identity checks Lanny91 2017-04-28 16:50:17 +01:00
  • 51d84ec057 Bugfixes in Wilson 5D sequential conserved current insertion Lanny91 2017-04-28 16:49:14 +01:00
  • db14fb30df Hadrons: overhaul of conserved current test Lanny91 2017-04-28 16:48:00 +01:00
  • b9356d3866 Added more complete test of sequential insertion of conserved current. Lanny91 2017-04-28 16:46:40 +01:00
  • 99a73f4287 Correcting the M and Mdag in the clover term Guido Cossu 2017-04-28 15:51:05 +01:00
  • f302eea91e SitePropagator redefined to be a scalar object in TYPE_ALIASES. Lanny91 2017-04-28 15:27:49 +01:00
  • 5553b8d2b8 Clover term compiles, not tested Guido Cossu 2017-04-28 15:23:34 +01:00
  • a6ccbbe108 Conserved current sequential source now registered properly and fixed module inputs. Lanny91 2017-04-28 10:43:47 +01:00
  • 3ac27e5596 QedFVol: remove unnecessary copies of free propagator from shifted sources in ScalarVP James Harrison 2017-04-27 14:17:50 +01:00
  • 99220f6531 Fixes and better timing Peter Boyle 2017-04-26 17:24:11 -04:00
  • e57eafe388 Fix to multinode code feature/half-prec-comms Peter Boyle 2017-04-26 14:46:52 -04:00
  • d2003f24f4 Corrected incorrect usage of ExtractSlice for conserved current code. Lanny91 2017-04-26 17:25:28 +01:00
  • 6299dd35f5 Hadrons: Added test of conserved current code. Tests Ward identities for conserved vector and partially conserved axial currents. Lanny91 2017-04-26 12:41:39 +01:00
  • a39daecb62 Removed make_5D const declaration to avoid compilation error Lanny91 2017-04-26 12:39:07 +01:00
  • 159770e21b Legal Banners added Lanny91 2017-04-26 09:32:57 +01:00
  • 2a6d093749 move the sudo: required to match locatoin on Guido's branch paboyle 2017-04-26 09:15:34 +01:00
  • c947947fad sudo required suggested by guido paboyle 2017-04-26 08:45:36 +01:00
  • f555b50547 Merge branch 'feature/half-prec-comms' into develop paboyle 2017-04-26 08:43:40 +01:00
  • 738c1a11c2 longer nloop paboyle 2017-04-26 08:43:20 +01:00
  • f8797e1e3e bug fix. works now and great face performance Peter Boyle 2017-04-26 03:14:02 -04:00
  • fd1eb7de13 Clean implementation of the exterior faces listing only those points on the boudary Peter Boyle 2017-04-26 02:34:52 -04:00
  • 2ce898efa3 Pretty code Peter Boyle 2017-04-26 02:34:25 -04:00
  • dc5a6404ea Hadrons: modules for testing conserved current contractions and sequential insertion. Lanny91 2017-04-25 22:08:33 +01:00
  • 44260643f6 First conserved current implementation for Wilson fermions only. Not implemented for Gparity or 5D-vectorised Wilson fermions. Lanny91 2017-04-25 18:00:24 +01:00
  • 1425afc72f Rare Kaon test fix Lanny91 2017-04-25 17:26:56 +01:00
  • bd466a55a8 QedFVol: remove charge dependence in chargedProp function of ScalarVP James Harrison 2017-04-25 10:04:03 +01:00
  • ab66bac4e6 Think I'm getting on top of the reduced cost exterior precomputed list of links paboyle 2017-04-25 08:50:26 +01:00
  • 56277a11c8 Build a list of whats on the surface paboyle 2017-04-24 17:06:15 +01:00
  • 752048f410 Merge branch 'develop' into feature/clover Guido Cossu 2017-04-24 14:41:20 +01:00
  • 916e9e1d3e Merge branch 'feature/half-prec-comms' of https://github.com/paboyle/Grid into feature/half-prec-comms paboyle 2017-04-24 10:39:19 +01:00
  • 5b55867a7a Slightly cheaper Ext assembly Peter Boyle 2017-04-24 05:36:11 -04:00
  • 3accb1ef89 Debugged assemply split phase with interior suppression Peter Boyle 2017-04-23 19:30:19 -04:00
  • e3d0e31525 Debugged assemply split phase with interior suppression Peter Boyle 2017-04-23 19:29:27 -04:00
  • 5812eb8a8c Partially fixed. But the comms-overlap does not work yet. Peter Boyle 2017-04-22 18:50:25 -04:00
  • 4dd3763294 Use OMP as much as possible paboyle 2017-04-22 20:35:20 +01:00
  • c429ace748 Cleaner OpenMP use paboyle 2017-04-22 20:28:42 +01:00
  • ac58565d0a Dangerous rewrite of the assembly. If I make a mistake the debug will be painful. paboyle 2017-04-22 19:31:04 +01:00
  • 3703b718aa Mark up a table if a given site only receives from itself; including MPI3 splitting info. paboyle 2017-04-22 19:28:37 +01:00
  • b722889234 Try a better load balancing loop paboyle 2017-04-22 19:27:41 +01:00
  • abba44a837 Hand unrolled for overlapped comms paboyle 2017-04-22 17:45:17 +01:00
  • f301be94ce Fixed paboyle 2017-04-22 17:42:31 +01:00
  • 1d1b225497 Hand unrolled Nc=3 kernels support split phase compute (on-node, off-node). Peter Boyle 2017-04-22 09:05:28 -04:00
  • 53a785a3dd Fixing the KNL compile Peter Boyle 2017-04-22 08:11:51 -04:00
  • 736bf3c866 Major rework of stencil. Half precision and MPI3 now working. paboyle 2017-04-22 11:33:50 +01:00
  • b9bbe5d188 L1p config bg/q paboyle 2017-04-22 11:33:09 +01:00
  • 3844bcf800 If no f16c instructions supported must use software half precision conversion. This will also become useful on BG/Q, so will move out from SSE4 into a general area. Lifted the Eigen half precision from web. Looks sensible, but not extensively regressed against the intrinsics implementation yet. paboyle 2017-04-20 15:30:52 +01:00
  • e1a2319d01 Simple compressor moved out of cshift into stencil paboyle 2017-04-20 13:18:15 +01:00
  • 180c732b4c Move compressors out of Cshift. Slice iterators would help paboyle 2017-04-20 13:17:55 +01:00
  • 957a706d0b Useful script paboyle 2017-04-20 13:17:44 +01:00
  • d2312e9874 Drop compressor entirely from Cshift to only Stencil. paboyle 2017-04-20 13:16:55 +01:00
  • fc4ab9ccd5 Working half precision comms paboyle 2017-04-20 11:20:26 +01:00
  • 4a340aa5ca Massive compressor rework to support reduced precision comms paboyle 2017-04-20 09:28:27 +01:00
  • 3b7de792d5 Type comparison in the traits work paboyle 2017-04-18 13:28:04 +01:00
  • 557c3fa109 Pretty change paboyle 2017-04-18 13:27:38 +01:00
  • ec18e9f7f6 Merge branch 'develop' into feature/half-prec-comms paboyle 2017-04-18 11:39:39 +01:00
  • a839d5bc55 Updated todo list paboyle 2017-04-18 11:22:17 +01:00
  • de41b84c5c Merge branch 'feature/normHP' into develop paboyle 2017-04-18 10:57:21 +01:00
  • 8e161152e4 MultiRHS solver improvements with slice operations moved into lattice and sped up. Block solver requires a lot of performance work. feature/normHP paboyle 2017-04-18 10:51:55 +01:00
  • 3141ebac10 MultiRHS working, starting to optimise. Block doesn't and I thought it already was; puzzled. paboyle 2017-04-17 10:50:19 +01:00
  • 7ede696126 Non compile of tests fixed paboyle 2017-04-16 23:40:00 +01:00
  • a07556dd5f Added back the convergence test from evecs of tridiagonal matrix. Bugfixes Chulwoo Jung 2017-04-15 09:32:15 -04:00
  • bf516c3b81 higher precision reduction variables in norm and inner product paboyle 2017-04-15 12:27:28 +01:00
  • 441a52ee5d First cut at higher precision reduction paboyle 2017-04-15 10:57:21 +01:00
  • a8db024c92 Cleaning up the dense matrix and lanczos sector paboyle 2017-04-15 08:54:11 +01:00
  • a9c22d5f43 Verbose removal paboyle 2017-04-14 14:38:49 +01:00
  • 3ca41458a3 Fix to no USE_FP16 case paboyle 2017-04-14 14:20:54 +01:00
  • 9e2d29c644 USE_FP16 macro paboyle 2017-04-14 14:17:14 +01:00
  • b694996302 adding comments Guido Cossu 2017-04-14 13:30:14 +01:00
  • 951be75292 Half precision conversion working on AVX512 now too Peter Boyle 2017-04-13 17:35:11 +01:00
  • c8e6f58e24 Fix typos in ScalarVP James Harrison 2017-04-13 17:04:37 +01:00
  • b9113ed310 Patches for knl Peter Boyle 2017-04-13 12:02:12 -04:00
  • 888988ad37 Merge branch 'feature/qed-fvol' of https://github.com/paboyle/Grid into feature/qed-fvol James Harrison 2017-04-13 15:54:40 +01:00
  • 1407418755 Old qed-fvol program build disabled Antonin Portelli 2017-04-13 15:32:30 +01:00
  • a6a0da873f Merge branch 'feature/hadrons' into feature/qed-fvol Antonin Portelli 2017-04-13 15:31:06 +01:00
  • 42fb49d3fd Merge branch 'develop' of https://github.com/paboyle/Grid into develop paboyle 2017-04-13 14:12:47 +01:00
  • 2a54c9aaab Merge branch 'feature/block-cg' into develop paboyle 2017-04-13 14:12:24 +01:00
  • 0957378679 Fixing conditional ugly way paboyle 2017-04-13 13:47:56 +01:00
  • 2ed6c76fc5 Getting multiline if then fi working paboyle 2017-04-13 13:43:13 +01:00
  • d3b9a7fa14 F16c apparently requires AVX, even if the 128 bit are used. Seems odd. paboyle 2017-04-13 13:19:11 +01:00
  • 75ea306ce9 Another try at travis paboyle 2017-04-13 13:05:32 +01:00
  • 4226c633c4 Default to FP16 off again paboyle 2017-04-13 12:51:39 +01:00