1
0
mirror of https://github.com/paboyle/Grid.git synced 2026-06-21 11:18:15 +01:00

Commit Graph

  • 4bc08ed995 Improved the prefetching when using cache blocking codes paboyle 2016-06-26 12:54:14 -07:00
  • b2933a0557 COntrol the prefetch strategy paboyle 2016-06-25 12:55:25 -07:00
  • db057cc276 Prefetch change paboyle 2016-06-25 12:54:50 -07:00
  • 22e88eaf54 Prefetch during save paboyle 2016-06-25 12:54:14 -07:00
  • 09fe3caebd Tweaks paboyle 2016-06-25 11:08:05 -07:00
  • 5e02392f9c Fixed compilation error for benchmark_dwf Guido Cossu 2016-06-20 12:30:51 +01:00
  • 17a8f51a9b update file lists paboyle 2016-06-19 11:59:10 -07:00
  • 1b7f88dd00 Enable reordering of the loops in the assembler for cache friendly. This gets in the way of L2 prefetching however. Do next next link in stencil prefetching. paboyle 2016-06-19 11:45:58 -07:00
  • d6737e4bd8 Travis fix for Linux clang builds portelli 2016-06-14 19:07:01 +01:00
  • 75fc295f6e Merge branch 'hadrons' into feature/hadrons portelli 2016-06-14 17:51:15 +01:00
  • d539888e57 Merge pull request #37 from rprollins/fix/mpi_communicator portelli 2016-06-14 17:25:40 +01:00
  • 86187d7cca Removed write to stdout in constructor for MPI CartesianCommunicator Richard Rollins 2016-06-14 15:34:20 +01:00
  • 87418e7df1 Slightly faster prefetching perf. paboyle 2016-06-13 02:32:52 -07:00
  • 55f65b81b5 Improvements to the assembler interface that let us move chunks of the site and s loop into the kernels. This will save on function call overhead and guarantee L2 prefetching strategy is right since OMP can't distribute the sub-chunks of work. paboyle 2016-06-09 01:12:36 -07:00
  • d9408893b3 Prefetching in the normal kernel implementation. Azusa Yamaguchi 2016-06-08 05:43:48 -07:00
  • 05acc22920 placeholder for non temporal loads optimisation paboyle 2016-06-07 13:18:21 -07:00
  • 8ac021de73 Added a test an fixed it for red black precon Ls innermost vectorised DWF paboyle 2016-06-07 13:16:56 -07:00
  • e503ef5590 Cleaned up paboyle 2016-06-07 00:11:36 +01:00
  • a7682b0060 Only instantiate the one routine to avoid duplicate symbol under g++5/MacOS paboyle 2016-06-06 23:48:21 +01:00
  • 0b731b5d80 Hadrons: genetic scheduler parameter fix portelli 2016-06-06 17:46:53 +01:00
  • 8e2078be71 Hadrons: environment with fully generic object store portelli 2016-06-06 17:45:37 +01:00
  • c5ab9f247f Merge branch 'master' into chulwoo-dec12-2015 chulwoo-dec12-2015 Chulwoo Jung 2016-06-06 10:57:27 -04:00
  • d4c9d71fc8 Merge branch 'master' of https://github.com/paboyle/Grid paboyle 2016-06-06 07:06:54 -07:00
  • 786ca52c43 Problems remain in the red black preconditioning of the Ls vectorisation paboyle 2016-06-06 07:05:51 -07:00
  • 048ac04abc Update Benchmark_dwf.cc Peter Boyle 2016-06-03 13:44:41 +01:00
  • f78d89bcbe Update Lebesgue.cc Peter Boyle 2016-06-03 13:33:42 +01:00
  • 53d06046b0 Compiling updates for KNL paboyle 2016-06-03 03:47:54 -07:00
  • 5d3a1a025d timers flag paboyle 2016-06-03 03:25:38 -07:00
  • 139cc5f1ae Large change with KNL preparation paboyle 2016-06-03 03:24:26 -07:00
  • 1826ed06a3 Merge branch 'master' into hadrons portelli 2016-05-27 16:50:31 +01:00
  • 1c0e922585 Merge pull request #35 from aportelli/master portelli 2016-05-27 16:49:13 +01:00
  • 2893a9b116 checking in before chaning to master Chulwoo Jung 2016-05-26 16:42:50 -04:00
  • 9d5f693cbe empty SIMD fix portelli 2016-05-24 10:56:27 +01:00
  • 5c90c3b457 Merge pull request #34 from aportelli/master Peter Boyle 2016-05-24 10:50:04 +01:00
  • 3ff96c502b Merge branch 'master' into hadrons portelli 2016-05-12 19:24:18 +01:00
  • 91e04056f9 fix of the empty SIMD portelli 2016-05-12 19:24:10 +01:00
  • 15a0908bfc Merge branch 'master' into hadrons portelli 2016-05-12 18:35:46 +01:00
  • 3789e3f31c additional fixed in slice functions portelli 2016-05-12 18:35:38 +01:00
  • bb2125962b Hadrons: finished implementation of 5D quarks portelli 2016-05-12 18:34:42 +01:00
  • 232fda5fe1 Hadrons: DWF action portelli 2016-05-12 18:34:10 +01:00
  • 2b31bf61ff Hadrons: message fix portelli 2016-05-12 18:33:49 +01:00
  • afe5a94745 Hadrons: getModule with upcast portelli 2016-05-12 18:33:36 +01:00
  • 7ae667c767 Hadrons: module template update portelli 2016-05-12 18:33:08 +01:00
  • 07f0b69784 Merge branch 'master' into hadrons portelli 2016-05-12 13:02:18 +01:00
  • 0c66719210 const fix in slice functions portelli 2016-05-12 13:01:35 +01:00
  • 5c06e89d69 Hadrons: code cleaning portelli 2016-05-12 12:49:49 +01:00
  • 3d75e0f0d1 Hadrons: MQuark fix portelli 2016-05-12 12:02:15 +01:00
  • 362f255100 Hadrons: module parameters can now be accessed from outside portelli 2016-05-12 11:59:28 +01:00
  • 3a5b5c8bec Save an old tar of tree paboyle 2016-05-12 03:20:17 -07:00
  • fdbe071213 space added paboyle 2016-05-12 02:59:51 -07:00
  • 3d78ed03ef Merge branch 'master' into hadrons portelli 2016-05-11 15:21:46 +01:00
  • 4bc21ec7cb thread CL argument fix portelli 2016-05-11 15:21:29 +01:00
  • e3083b6dfc Merge commit 'ab894186589224d570e0ecef8eea06443194a8ab' portelli 2016-05-11 15:20:41 +01:00
  • ab89418658 Precision change going in; useful for mixed precision algorithms for example. paboyle 2016-05-11 15:18:47 +01:00
  • 28cd99882c Subslicing paboyle 2016-05-11 15:06:54 +01:00
  • 835003b3c5 Hadrons: removed useless gauge global parameters portelli 2016-05-11 15:01:52 +01:00
  • 328d213c9e Hadrons: FS case sensitivity fix portelli 2016-05-11 14:44:14 +01:00
  • aceaee774c ExtractSlice / InsertSlice for lower dimensional lattices where the lattice is not distributed in the orthogonal direction. Useful for fermion 4d/5d etc.. paboyle 2016-05-11 14:12:02 +01:00
  • 56a8d7a5bc Hadrons: build system fix portelli 2016-05-11 10:27:14 +01:00
  • 78198d1b04 Hadrons: size fix for module graph with one vertex portelli 2016-05-10 20:13:28 +01:00
  • 84fa2bdce6 Hadrons: modules moved in their own directory & utility script to add new modules portelli 2016-05-10 20:12:48 +01:00
  • 29dfe99e7c Hadrons: more scheduler optimizations portelli 2016-05-10 19:19:38 +01:00
  • d604580e5a Hadrons: all objects/modules mapped to an integer address system to remove string operations from scheduling portelli 2016-05-10 19:07:41 +01:00
  • 7dfdc9baa0 Hadrons: lattice dynamic cast fix portelli 2016-05-10 10:41:20 +01:00
  • 9e986654e6 Hadrons: first version of the genetic scheduler portelli 2016-05-09 14:49:06 +01:00
  • df3fbc477e Hadrons: code cleaning portelli 2016-05-07 13:26:56 -07:00
  • bb580ae077 Hadrons: significant overhaul of the object registration system, previous version didn't allow dry runs portelli 2016-05-07 13:19:38 -07:00
  • 2c226753ab Hadrons: comments on graph theory algorithm complexity portelli 2016-05-06 06:35:11 -07:00
  • ea0cea668e Hadrons: minor code cleaning portelli 2016-05-05 16:13:14 -07:00
  • f8f9fd6f22 Merge pull request #33 from aportelli/master Peter Boyle 2016-05-05 22:57:13 +01:00
  • 75cd72a421 Hadrons: memory management for fermion matrices, dynamic ownership in garbage collector portelli 2016-05-04 19:11:03 -07:00
  • cbe52b0659 Hadrons: debug message removed portelli 2016-05-04 12:19:58 -07:00
  • 3aa6463ede Hadrons: general lattice store & a lot of code cleaning portelli 2016-05-04 12:17:27 -07:00
  • 312637e5fb Merge branch 'master' into hadrons portelli 2016-05-04 12:16:18 -07:00
  • 101aa769eb LatticeBase contain the grid pointer and a virtual destructor to allow polymorphic lattice pointers portelli 2016-05-04 12:15:31 -07:00
  • 0bf99bfde5 log polish portelli 2016-05-04 12:14:49 -07:00
  • 64bf6fe54e macro to dump NERSC header to a stream portelli 2016-05-04 12:14:38 -07:00
  • 798d8f7340 Hadrons: Modules: better log messages portelli 2016-05-03 18:17:58 -07:00
  • ba878724ce Hadrons: sources are now independent modules portelli 2016-05-03 18:17:28 -07:00
  • b865dd9da8 Hadrons: solver renaming portelli 2016-05-03 18:16:57 -07:00
  • 8b313a35ac Hadrons: random and NERSC gauge configurations portelli 2016-05-03 17:07:00 -07:00
  • 02ec23cdad Hadrons: Fermion actions and gauge fields are modules now portelli 2016-05-03 16:30:29 -07:00
  • 1161d566b9 minor code cleaning portelli 2016-05-02 19:32:11 -07:00
  • 6e83b6a203 Hadrons: namespace reorganisation, now everything is in Grid::Hadrons, the 'using Grid::operator<<' statement is used to prevent a very nasty compilation error with GCC. portelli 2016-05-02 19:31:21 -07:00
  • 48fcc34d72 CMeson: first implementation, still need proper output portelli 2016-05-01 18:31:40 -07:00
  • d08d93c44c Merge branch 'master' into hadrons portelli 2016-05-01 18:30:44 -07:00
  • c698b16d75 function to generate Chroma-style gamma matrix products portelli 2016-05-01 18:30:35 -07:00
  • c4c89336fe SliceSum: shutting down warning about non-threaded code for now portelli 2016-05-01 18:29:57 -07:00
  • fa59789580 ConjugateGradient: cleaner output portelli 2016-05-01 18:29:20 -07:00
  • 0ab10cdedb Merge branch 'master' into hadrons portelli 2016-05-01 16:08:05 -07:00
  • 92c2c7d3b5 SchurRedBlackDiagMooeeSolve: fix: guess was not initialised from input portelli 2016-05-01 16:07:55 -07:00
  • e99ce0875f directly exit when using '--help' option portelli 2016-05-01 16:05:16 -07:00
  • 22653edf12 Merge branch 'master' into hadrons portelli 2016-05-01 15:55:58 -07:00
  • cc1d9eb05b Merge commit '999b3a2e26bdd8300d389699dd299e7e5d951af6' portelli 2016-05-01 15:55:22 -07:00
  • 12d2a95846 Merge branch 'master' into hadrons portelli 2016-05-01 15:05:02 -07:00
  • 57c027fea2 Travis update portelli 2016-05-01 15:04:52 -07:00
  • 207dc439a7 Travis debug portelli 2016-05-01 15:00:35 -07:00
  • 978cf52f6b Merge branch 'master' into hadrons portelli 2016-05-01 14:53:38 -07:00
  • 77ef0bba48 Travis update portelli 2016-05-01 14:53:28 -07:00
  • 63b730de80 Hadrons: for the moment, test with unit gauge portelli 2016-05-01 14:50:57 -07:00