feat: Build NCCL from source for more up-to-date versions#35
Draft
Eta0 wants to merge 2 commits into
Draft
Conversation
wbrown
approved these changes
Jul 28, 2025
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Build NCCL from source
This change switches the installation of
libnccl2andlibnccl-devfromapt-getto a source build.The prebuilt NCCL distributions available for an
apt-getinstall are not updated consistently across all OS version × CUDA version pairs, so this change ensures all of our images can use new NCCL releases as they come out.This change additionally updates the NCCL version to v2.21.5 on all of our current build targets.
This build supports the following (configurable) list of compute architectures by default: 7.0, 7.5, 8.0, 8.6, 8.9, 9.0+PTX (matching our ml-containers PyTorch builds). This is slightly smaller than the default architecture support list, so the binaries are a bit smaller than the prebuilt distribution as well.