Commits · cac5c3f5f75b3afe5e28a3a00aa47bb9de7bcb17 · SWIFT / SWIFTsim · GitLab

Snippets Groups Projects

Aug 11, 2017
- Use an informed guess about how many tasks to use per cell · 68e597c3
  Peter W. Draper authored 7 years ago
  68e597c3
Aug 04, 2017
- Make the expected number of tasks per cell a parameter. · 093ec68d
  Peter W. Draper authored 7 years ago
  093ec68d
Aug 03, 2017
- engine_launch is only ever called with all threads, remove the option to limit... · 091462a9
  Pedro Gonnet authored 7 years ago
  091462a9
- use the new barrier. · a473067b
  Pedro Gonnet authored 7 years ago
  
  a473067b
Jul 31, 2017
- Recover more missed updates from original threadpool_task_plots · d439130a
  Peter W. Draper authored 7 years ago
  
  d439130a
Jul 30, 2017
- engine_launch is only ever called with all threads, remove the option to limit... · be037f3e
  Pedro Gonnet authored 7 years ago
  be037f3e
- use the new barrier. · cd6bd663
  Pedro Gonnet authored 7 years ago
  
  cd6bd663
Jul 28, 2017
- Revert "Merge branch 'threadpool_task_plots' into 'master' · e3309772
  Peter W. Draper authored 7 years ago
  e3309772
Jul 27, 2017
- set engine_taskreweight to 1 so that it never happens. · 0de74269
  Pedro Gonnet authored 7 years ago
  
  0de74269
Jul 13, 2017
- Move check to before we read the ICs and sort out issue with the result from dirname() · f0639090
  Peter W. Draper authored 7 years ago
  f0639090
Jun 23, 2017
- Add missing mpoles policy name and fix maximum policy mask · a559a3f7
  Peter W. Draper authored 7 years ago
  a559a3f7
- Updated the cell sanitization tool to give more sensible values to the extreme h outliers. · f8be5419
  Matthieu Schaller authored 7 years ago
  
  f8be5419
May 24, 2017
- add non-implicit ghosts below the ghost_in/ghost_out tasks that do the actual work. · b7636553
  Pedro Gonnet authored 8 years ago
  
  b7636553
Apr 20, 2017
- Added the option to reconstruct all the multipoles every time-step (expensive) · 107aa7e5
  Matthieu Schaller authored 8 years ago
  
  107aa7e5
Apr 01, 2017
- Drift all the top-level multipoles to current time before launching the tasks. · 2287f0e6
  Matthieu Schaller authored 8 years ago
  
  2287f0e6
Mar 22, 2017
- Introduce a collectgroup API to handle the group of values we want to MPI reduce at the same time · c0d331e7
  Peter W. Draper authored 8 years ago
  c0d331e7
Mar 15, 2017
- Move logic about whether to trigger a repartition into function and remove the · fc3e4d3e
  Peter W. Draper authored 8 years ago
  fc3e4d3e
Mar 13, 2017
- Stop repartitioning in the step after a repartition, that makes no sense and... · d76b4aa1
  Peter W. Draper authored 8 years ago
  d76b4aa1
Mar 10, 2017
- Extend possible schemes to include a number of steps as well as a · bd803b48
  Peter W. Draper authored 8 years ago
  bd803b48
Mar 04, 2017
- Simplified the calculation of active/inactive by pre-computing the largest active time-bin · b562e3d8
  Matthieu Schaller authored 8 years ago
  
  b562e3d8
Feb 26, 2017
- Drift before dumping statistics. · ff45a6ac
  Matthieu Schaller authored 8 years ago
  
  ff45a6ac
Feb 22, 2017
- Properties of the self-gravity scheme are now read from the YAML file. They... · 7d79c8be
  Matthieu Schaller authored 8 years ago
  7d79c8be
Feb 21, 2017
- Re-factored the UnitSytem to use the same naming convention as the rest of the code. · 039540fd
  Matthieu Schaller authored 8 years ago
  
  039540fd
Feb 17, 2017
- Code formatting and doxygen documentation · dc1ff243
  Matthieu Schaller authored 8 years ago
  
  dc1ff243
- Move the repartitionning decision to engine_step() · f595959b
  Matthieu Schaller authored 8 years ago
  
  f595959b
Feb 12, 2017
- Time-step re-ordered · 4a6150af
  Matthieu Schaller authored 8 years ago
  
  4a6150af
Feb 10, 2017
- Re-arranged the time-step sequence. Still some debugging checks in place. · 8f560a8a
  Matthieu Schaller authored 8 years ago
  
  8f560a8a
Jan 24, 2017
- Star particles can now be used over MPI. · bb9bde8e
  Matthieu Schaller authored 8 years ago
  
  bb9bde8e
- Added star formation as an engine policy · db251a47
  Matthieu Schaller authored 8 years ago
  
  db251a47
Jan 23, 2017
- Add the s-particles to the time integration tasks. · cdad747a
  Matthieu Schaller authored 8 years ago
  
  cdad747a
- After repartitioning don't check if all cells have been drifted · 43d8be2d
  Peter W. Draper authored 8 years ago
  43d8be2d
Dec 24, 2016
- Started moving towards 50 time bins. Time-step of size 0 is assigned. · 5d91cae2
  Matthieu Schaller authored 8 years ago
  
  5d91cae2
Dec 22, 2016
- Need to unskip before rebuild, except when following a repartition, so make that work · 598ae8c0
  Peter W. Draper authored 8 years ago
  
  598ae8c0
Dec 09, 2016
- Cleaning up · b3dbade2
  Matthieu Schaller authored 8 years ago
  
  b3dbade2
Dec 08, 2016
- The step tic/toc is now only used in the thread task dumps · 30bb8907
  Peter W. Draper authored 8 years ago
  
  30bb8907
- Switch to using CPU time to estimate the fraction difference in runtimes between ranks · 43df5f7b
  Peter W. Draper authored 8 years ago
  43df5f7b
Dec 06, 2016
- Make fraction time difference that triggers a repartition a parameter · 48c45be5
  Peter W. Draper authored 8 years ago
  
  48c45be5
- Rephrased the logic in engine_step() in terms of task-unskipping and cell-drifting · e1b61cbe
  Matthieu Schaller authored 8 years ago
  
  e1b61cbe
Nov 28, 2016
- Particles only carry their time bin. · 8bcd1ff4
  Matthieu Schaller authored 8 years ago
  
  8bcd1ff4
Oct 30, 2016
- Remove the fixdt policy. fixdt can be achieved by setting dt_min and dt_max to the same value. · 74606c8b
  Matthieu Schaller authored 8 years ago
  
  74606c8b