Skip to content

Refactor/gen on worker try again#1703

Merged
jlnav merged 60 commits intodevelopfrom
refactor/gen_on_worker_try_again
Apr 13, 2026
Merged

Refactor/gen on worker try again#1703
jlnav merged 60 commits intodevelopfrom
refactor/gen_on_worker_try_again

Conversation

@jlnav
Copy link
Copy Markdown
Member

@jlnav jlnav commented Mar 27, 2026

  • Remove gen_on_manager and replace with gen_on_worker. Remove docs mentions of gen_on_manager.

  • Remove zero_resource_workers.

  • Misc typing adjustments and fixes.

  • Various corresponding indexing fixes for the gen to get assigned GPUS/procs separately.

    • get_index_list needed a tricky adjust. First element is None since worker 0's resources are handled separately.
  • Adjustments to the tests. Some tests like the profile and splitcomm test are flakey without gen_on_worker.

  • Docs and tutorials updates.

  • Some forces-building adjusts for macOS compiling.

  • TODO: Undo overkill changes to test_GPU_gen_resources. Obviously hot-swapping hardware mid-loop shouldn't fix anything

Addresses #1426

jlnav added 30 commits May 2, 2025 09:22
…kers (with gen_on_worker defaulting to False) clear to resources. other temporary debug adjusts
…ors, regarding the default presence of the zeroth worker. assistance from gemini/claude for fixing resource math in various tests.
… worker indexes and zrws. mypy types. vibe-coded new test cases for test_mpi_runners_zrw_subnode_uneven
jlnav added 17 commits March 26, 2026 16:23
… procs matched to GPUs we really ought to assign sims to those resource sets first. Only if no GPU rsets exist do we fallback to non-gpu sets, producing a correct wait upon allocation. Thanks Claude.
…tent gen is always available to receive gen work. so to prevent worker 0 from generating extra, set sim max. another way to prevent extra non-persistent gen calls is set gen_on_worker=True
@codecov
Copy link
Copy Markdown

codecov bot commented Apr 8, 2026

Codecov Report

❌ Patch coverage is 91.76471% with 7 lines in your changes missing coverage. Please review.
✅ Project coverage is 78.02%. Comparing base (3d20f6f) to head (92d032e).
⚠️ Report is 21 commits behind head on develop.

Files with missing lines Patch % Lines
libensemble/ensemble.py 85.71% 0 Missing and 2 partials ⚠️
libensemble/libE.py 81.81% 1 Missing and 1 partial ⚠️
libensemble/manager.py 95.00% 1 Missing ⚠️
libensemble/resources/rset_resources.py 66.66% 1 Missing ⚠️
libensemble/worker.py 90.90% 1 Missing ⚠️

❗ There is a different number of reports uploaded between BASE (3d20f6f) and HEAD (92d032e). Click for more details.

HEAD has 13 uploads less than BASE
Flag BASE (3d20f6f) HEAD (92d032e)
22 9
Additional details and impacted files
@@             Coverage Diff              @@
##           develop    #1703       +/-   ##
============================================
- Coverage    92.84%   78.02%   -14.82%     
============================================
  Files           81       81               
  Lines         8218     8210        -8     
  Branches      1259     1259               
============================================
- Hits          7630     6406     -1224     
- Misses         387     1563     +1176     
- Partials       201      241       +40     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

@jlnav jlnav requested review from jmlarson1 and shuds13 April 9, 2026 16:26
@jlnav jlnav merged commit 326855e into develop Apr 13, 2026
11 of 12 checks passed
@jlnav jlnav deleted the refactor/gen_on_worker_try_again branch April 13, 2026 13:28
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants