[EP] config rdma #619
Open
[EP] config rdma #619
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
) * Fix readme. * Remove hardcode macro. * WIP. IBRC for p2p. * USE_QPEX. * Nits. * Fix qp init bug. * IBRC now is working on new p2p arch. * Format code. * Inlining small metadata. * Remove unused macro. * Rename macro. * Rename efa folder to rdma. * Refactor p2p * Format code. * Nits. * Nits. * Fix Makefile. * Nits. * Nits. * Nits. * fix p2p on efa * deprecate p2p chunk size and entropy, add UCCL_IB_GID_INDEX * change kNICContextNumber to 4 to allow 4 EFA nics per GPU * renaming a bit --------- Co-authored-by: YangZhou1997 <[email protected]>
Co-authored-by: Ziming Mao <[email protected]>
MaoZiming
reviewed
Jan 6, 2026
| // Add self-ranks, sub other ranks | ||
| if (thread_id < kNumRanks) { | ||
| atomicAdd_system(barrier_signal_ptrs[rank] + thread_id, FINISHED_SUM_TAG); | ||
| memory_fence(); |
Member
There was a problem hiding this comment.
Is the memory_fence() needed here?
Member
Author
There was a problem hiding this comment.
I followed rocm/deepep https://github.com/ROCm/DeepEP/blob/main/csrc/kernels/utils.cuh#L792
…m/uccl-project/uccl into ep-debug-amd-mem-consistency-yang
* add UdsExchanger for Ipc * IpcEndpoint * gpu_id->local_rank; rank->glocal_rank * communcator test with multi processes * format * add ipc cache
* add vllm launch scripts * pass on vLLM setup * add tiemout * fixing several issues * nits * nits * nits * refining readme * nits * add DG_JIT_CACHE_DIR * nits * debug vvlm launch * dp=2, tp=8 works for ll mode * continue debugging * remove conditional * remove print * remove print * newline * add readme * fomrat --------- Co-authored-by: YangZhou1997 <[email protected]> Co-authored-by: MaoZiming <[email protected]>
…ebug-amd-mem-consistency-yang
…ebug-amd-mem-consistency-yang
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.
Description
Please include a summary of the changes and the related issue.
Fixes # (issue)
Type of Change
How Has This Been Tested?
Include any tests here.
Checklist
format.sh.build_and_install.shto verify compilation.