Move horovod and sok init functions to user code#1135
Closed
edknv wants to merge 3 commits intoNVIDIA-Merlin:mainfrom
Closed
Move horovod and sok init functions to user code#1135edknv wants to merge 3 commits intoNVIDIA-Merlin:mainfrom
edknv wants to merge 3 commits intoNVIDIA-Merlin:mainfrom
Conversation
Documentation preview |
edknv
changed the title
Contributor
Author
|
Ci is unblocked for now in #1136. Will close this PR and revisit if necessary. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Goals ⚽
hvd.init()andsok.init()and have the users run them.singlegpuandmultigpumarkers.Implementation Details 🚧
In some cases when horovod is not built correctly or not configured correctly with MPI,
hvd.init()orsok.init()will throw an MPI_LIB errors on the C-level, which we cannot catch from Python. This happened in #1134 for example.Following @oliverholworthy's proposal, this PR leaves
hvd.init()and/orsok.init()up to the user.We still need to initialize them to run unit tests but incorrect installation in the ci-runner will still block the CI, so this PR also introduces
singlegpuandmultigpupytest markers to skip the horovod tests in the single-gpu case, similarly to NVIDIA-Merlin/Merlin#999.