basic CUDA <> CPU or CUDA <> CUDA rdma Support #372
Conversation
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
Summary: exposes basic cuda bindings to monarch for rdma support Differential Revision: D77404103
Summary: expose rdmacore bindings; including basic ibv verbs along with mlx5dv prodivers to monarch for rdma. Differential Revision: D77408652
Summary: RDMA support for CUDA <> CUDA and CUDA <> CPU comms Key changes - using cuda apis we can detect if a given pointer is mapped to a cuda device, or cpu. - if data pointer is cuda, the code leverages dma registration to register with NIC; we are able to avoid directly passing with cuda allocation handles using cuMemGetHandleForAddressRange. - if data pointer is cpu, we use standard ibv mr; note I transitioned to using standard registration, not entire memory space (security concern raised by mariusae) - Refactored test infra to support named NIC devices, and different compute (cuda:X or cpu) This implementation is relatively naive, and I will iterate accordingly. To Do: add unit test for cuda/cuda Differential Revision: D77408653
facebook-github-bot
added
the
CLA Signed
|
This pull request was exported from Phabricator. Differential Revision: D77408653 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Summary:
RDMA support for CUDA <> CUDA and CUDA <> CPU comms
Key changes
This implementation is relatively naive, and I will iterate accordingly.
To Do: add unit test for cuda/cuda
Differential Revision: D77408653