Commit Graph

11889 Commits

Author SHA1 Message Date
Stan Moore 067119f6c6 Adding missing friend statement to pair_lj_class2_coul_cut_kokkos 2017-02-02 15:21:30 -07:00
Stan Moore 1834a5e46c Fixing more Kokkos per-atom and fdotr issues 2017-02-02 15:21:21 -07:00
Stan Moore 6a4918b39a Fixing typo in pair_buck_coul_cut_kokkos 2017-02-02 15:21:05 -07:00
Stan Moore 5da0d39392 Fixing fdotr in pair_buck_coul_cut_kokkos 2017-02-02 13:35:51 -07:00
Stan Moore 6f92429602 Fixing per-atom ev issue 2017-02-02 13:34:27 -07:00
Stan Moore 38e0e4bb69 Add missing typedef in Kokkos pair styles 2017-02-02 13:24:05 -07:00
Stan Moore daf9f95381 Fixing Kokkos per-atom e/v issue 2017-02-02 13:09:52 -07:00
sjplimp 997142a4c1 Merge pull request #364 from stanmoore1/kk_triclinic_neighlist
Add triclinic neighbor list support to Kokkos
2017-01-30 07:27:02 -07:00
sjplimp 033b07fdb7 Merge pull request #363 from ibaned/obey-datamask
Fix GPU sync bugs
2017-01-30 07:26:49 -07:00
Stan Moore 59f4a77dd5 Whitespace change to npair_kokkos 2017-01-27 15:17:39 -07:00
Stan Moore 579cc6d7aa More tweaks to npair_kokkos for triclinic 2017-01-27 15:13:37 -07:00
Stan Moore 5afd3e995b Adding support to npair_kokkos for triclinic-newton-on neighborlists 2017-01-27 14:18:01 -07:00
Dan Ibanez 2a6f5e651c more preference of datamask over custom sync
see commit 09fc8b0 for details on why
2017-01-27 09:35:55 -07:00
Dan Ibanez 09fc8b0bd7 kspace & dihedral can't do their own sync/modify
because the verlet_kokkos system has
a "clever" optimization which will
alter the datamasks before calling sync/modify,
so the datamask framework must be
strictly obeyed for GPU correctness.
(the optimization is to concurrently
compute forces on the host and GPU,
and add them up at the end of an iteration.
calling your own sync will overwrite
the partial GPU forces with the
partial host forces).
2017-01-27 08:39:55 -07:00
Steve Plimpton f8506fee23 sync GHub with SVN 2017-01-26 14:06:43 -07:00
sjplimp 18e5584311 Merge pull request #354 from stanmoore1/kokkos_bugfixes
Kokkos bugfixes
2017-01-26 13:51:47 -07:00
sjplimp 851f80464f Merge pull request #361 from akohlmey/user-omp-fix-per-atom-data
fix USER-OMP bug on per-atom data with hybrid styles
2017-01-26 13:50:13 -07:00
sjplimp 5971d4c994 Merge pull request #358 from ibaned/warnings
warning fixes (Kokkos+CUDA)
2017-01-26 13:49:56 -07:00
sjplimp 868d95f0a5 Merge pull request #352 from akohlmey/fix-skip-with-ghost-issue
Fix skip with ghost issue
2017-01-26 13:47:12 -07:00
sjplimp a5ff35435a Merge pull request #351 from timattox/USER-DPD_pair_exp6_bugfix
USER-DPD: Possible uninitialized variable in pair_exp6_rx.cpp bugfix.
2017-01-26 13:45:45 -07:00
Axel Kohlmeyer 8b7bd9d88e fix bug where per atom data for USER-OMP was reducing the wrong arrays with hybrid styles 2017-01-26 14:59:10 -05:00
Dan Ibanez 672bbbe494 add more missing KOKKOS_INLINE_FUNCTION attributes 2017-01-25 16:03:11 -07:00
Dan Ibanez 03c9c46533 add missing KOKKOS_INLINE_FUNCTION attributes 2017-01-25 15:49:05 -07:00
Dan Ibanez e992bfe510 remove unused variable 2017-01-25 15:40:52 -07:00
Dan Ibanez 053ee54a27 remove unused variable 2017-01-25 15:38:41 -07:00
Dan Ibanez 1074c6734b add missing return keywords 2017-01-25 15:37:27 -07:00
Dan Ibanez 60b48c9d66 add missing KOKKOS_INLINE_FUNCTION attributes
this structure gets put inside a DualView,
so these members need to be able to execute
on the GPU
2017-01-25 15:36:24 -07:00
Dan Ibanez 3d40b51708 remove unused variable 2017-01-25 15:24:52 -07:00
Dan Ibanez effbe18c46 fix domain boundary indexing
the compiler pointed out that
boundary[2][2] doesn't exist.
If I understand this correctly,
those checks should be against
boundary[*][0].
2017-01-25 15:24:01 -07:00
Dan Ibanez 6328beb7d7 fix double-return warning
this #ifdef adds a return statement
for little endian machines, but leaves
the old one, which the compiler comlains
is unreachable. this commit combines
the conditionals so we can use #else
2017-01-25 15:22:42 -07:00
Stan Moore 26c8d3d98f Fixing GPU memory issue in fix_langevin_kokkos 2017-01-25 12:53:55 -07:00
Stan Moore 73177d650d Fixing GPU memory issue in domain_kokkos 2017-01-25 11:18:03 -07:00
Axel Kohlmeyer b5cb74bd33 skip list build is compatible with NP_GHOST 2017-01-23 19:21:48 -05:00
Axel Kohlmeyer 31976d1dee skip list definition was missing NP_HALFFULL flag 2017-01-23 19:20:05 -05:00
Christopher Stone c8260af37c Possible uninitialized variable in USER-DPD/pair_exp6_rx.cpp bugfix.
Added explicit initialization (to zero) for several variables inside the
inner j-loop to avoid using them uninitialized or from prior iterations
within rmOldij_12 == 0.
2017-01-23 13:34:51 -05:00
Steve Plimpton caea8973a3 add neighbor list kind output to screen 2017-01-20 13:24:09 -07:00
sjplimp aa0ad9b483 Merge pull request #349 from akohlmey/collected-small-fixes
collected fixes and improvements
2017-01-20 13:19:43 -07:00
sjplimp 5d0e4e1ba9 Merge pull request #346 from stanmoore1/kokkos_fixes
Kokkos fixes
2017-01-20 13:15:16 -07:00
sjplimp f8d3c4c740 Merge pull request #345 from timattox/USER-DPD_another_zero_compute
USER-DPD another zero compute optimization
2017-01-20 13:14:59 -07:00
Axel Kohlmeyer e6996121d1 remove dead code 2017-01-20 14:30:46 -05:00
Axel Kohlmeyer fbfb1df5eb fix typo causing wrong neighbor list copy selections 2017-01-19 20:47:10 -05:00
Steve Plimpton 9a299875da simplified neighbor list copying to avoid possible same-timestep re-build issues 2017-01-19 17:01:15 -07:00
Stan Moore fc94f1bd18 Fixing GPU memory issues in Kokkos 2017-01-19 12:14:25 -07:00
Stan Moore 5ce8e2fbae Fixing GPU memory issue in modify_kokkos, need to cherry pick back to Master 2017-01-19 12:13:48 -07:00
Tim Mattox f6cd98636b USER-DPD: Also apply "check if a0 is zero" optimization to pair_dpd_fdt
This relates to commit 4eb08a5822 that was applied to pair_dpd_fdt_energy
2017-01-18 16:17:11 -05:00
Tim Mattox 05cafb716f USER-DPD: cleanup initialization of splitFDT_flag in pair_dpd_fdt.cpp 2017-01-18 15:51:50 -05:00
sjplimp 3af4b3c28c Merge pull request #337 from ohenrich/user-cgdna
Added source code and documentation for USER-CGDNA
2017-01-18 11:31:35 -07:00
sjplimp 7fc0970587 Merge pull request #344 from timattox/USER-DPD_zero_compute
USER-DPD: Skip a0*stuff computations, if a0 was set to zero in pair_coeff
2017-01-18 11:31:14 -07:00
sjplimp 93262b52b4 Merge pull request #343 from timattox/USER-DPD_bugfix_molecule
USER-DPD: bugfix for a segfault when using MOLECULE and DPD together.
2017-01-18 11:30:58 -07:00
Tim Mattox 4eb08a5822 USER-DPD: Skip a0*stuff computations, if a0 was set to zero in pair_coeff.
This saves around 10% of the runtime for many of our tests using SSA.
2017-01-17 15:55:39 -05:00