ESCOMP / ESCOMP/CMEPS

Problems with ESMF managed threading in CESM

Closed
#701 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
Fortran
Stars
35
Forks
106
Avg merge
6d 5h
Merged PRs (30d)
4

Description

In CESM, ESMF_AWARE_THREADING (ESMF's managed threading) results in hangs because of issues with how PEs that are threads are treated like full PETs in CMEPS & NUOPC at the moment, but they can't do things like query the component VM. Also, connectors built on PE lists (vs PET lists) would hang during RouteHandle creation.

Contributor guide

No contributing guide indexed for this repository

Research direction

Start by reproducing the hang with ESMF_AWARE_THREADING enabled in CESM, then trace how CMEPS and NUOPC treat thread PEs versus PETs and how PE-list connectors create RouteHandles. Done means managed threading no longer hangs, component VM queries work for threaded PEs, and PE-list connector creation completes.

Written by the indexing model from the issue text.

Assessment

Tech stack
fortran
Domain
distributed-systems, hpc
Issue type
Bug
Difficulty
5/5
Estimated time
Over a week
Activity status
Active
Clarity
Needs clarification
Newbie friendliness
30/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.