Skip to content

OpenMPI + CUDA Compatiblity#2279

Open
JuanPedroGHM wants to merge 4 commits into
mainfrom
fix/openmpi-gpu-compatibility-2.0
Open

OpenMPI + CUDA Compatiblity#2279
JuanPedroGHM wants to merge 4 commits into
mainfrom
fix/openmpi-gpu-compatibility-2.0

Conversation

@JuanPedroGHM

@JuanPedroGHM JuanPedroGHM commented May 6, 2026

Copy link
Copy Markdown
Member

Introduces an incompatibility list for different communication backends, in order to safely enable GPU compatible MPI communication.

Changes

  • Incompatibility list in configuration file.
  • Incompatibility check in _moveToCompDevice

Does this affect performance

  • Some operations (Bcast and Allgatherv in resplit_(None)) have shown massive slowdowns when running the tests. We might want to add certain functions to the incompatibility list for performance reasons.

@github-project-automation github-project-automation Bot moved this to Todo in Roadmap May 6, 2026
@JuanPedroGHM JuanPedroGHM linked an issue May 11, 2026 that may be closed by this pull request
@codecov

codecov Bot commented Jun 1, 2026

Copy link
Copy Markdown

Codecov Report

❌ Patch coverage is 78.72340% with 10 lines in your changes missing coverage. Please review.
✅ Project coverage is 91.67%. Comparing base (86eb546) to head (f7a75f8).
⚠️ Report is 13 commits behind head on main.
✅ All tests successful. No failed tests found.

Files with missing lines Patch % Lines
heat/core/_config.py 61.53% 5 Missing ⚠️
heat/core/communication.py 85.29% 5 Missing ⚠️
Additional details and impacted files
@@            Coverage Diff             @@
##             main    #2279      +/-   ##
==========================================
- Coverage   91.71%   91.67%   -0.04%     
==========================================
  Files          87       87              
  Lines       14284    14310      +26     
==========================================
+ Hits        13100    13119      +19     
- Misses       1184     1191       +7     
Flag Coverage Δ
unit 91.67% <78.72%> (-0.04%) ⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

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

@JuanPedroGHM JuanPedroGHM marked this pull request as ready for review June 3, 2026 08:25
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

Status: Todo

Development

Successfully merging this pull request may close these issues.

[Bug]: OpenMPI and CUDA buffers

1 participant