Skip to content
Snippets Groups Projects

Improve unskip by escaping early

Merged Improve unskip by escaping early
improve_unskip into master
1 unresolved thread
Merged Matthieu Schaller requested to merge improve_unskip into master
1 unresolved thread

When unskipping tasks, abort if a cell has no part or no gpart.

(note that the changes to cell_make_multipoles() are mixed in by mistake but are in a currently unused function)

Merge request reports

Merged by avatar (May 28, 2025 11:03am UTC)

Loading

Activity

Filter activity
  • Approvals
  • Assignees & reviewers
  • Comments (from bots)
  • Comments (from users)
  • Commits & branches
  • Edits
  • Labels
  • Lock status
  • Mentions
  • Merge request status
  • Tracking
src/cell.c
1518 1535 }
1519 1536 }
1520 1537
1521 /* Otherwise, pair interation, recurse? */
1522 else if (cell_can_recurse_in_pair_task(ci) &&
1523 cell_can_recurse_in_pair_task(cj)) {
1538 /* Otherwise, pair interation */
1539 else {
1540
1541 /* Should we even bother? */
1542 if (!cell_is_active_hydro(ci, e) && !cell_is_active_hydro(cj, e)) return;
  • Matthieu Schaller
    Matthieu Schaller @matthieu
    Author Owner

    Did you have further comments on this?

  • Pedro Gonnet
    Pedro Gonnet @nnrw56
    Developer

    LGTM, do you have any numbers showing whether this has a significant effect?

  • Matthieu Schaller
    Matthieu Schaller @matthieu
    Author Owner

    Not yet because our standard benchmark does not mix task types (only hydro and external gravity) but will make some more plots soon, once some changes to the gravity fixes are pushed in.

  • Pedro Gonnet
    Pedro Gonnet @nnrw56
    Developer

    OK, no need to wait for the benchmarks before merging, this is a reasonable change as-is.

  • Matthieu Schaller merged

    merged

  • Matthieu Schaller mentioned in commit db8c453d

    mentioned in commit db8c453d

    • Please register or sign in to reply