UCT/GDA: Use wqe_idx for rc_gda progress #10928
Merged
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
Artemy-Mellanox
force-pushed
the
topic/comp_wqe_idx
branch
from
aadb4d2 to
fa5d4d4
Compare
yosefe
reviewed
Sep 30, 2025
| if (req != nullptr) { | ||
| comp = &req->comp; | ||
| req->device_ep = device_ep; | ||
| uct_device_completion_init(comp); |
There was a problem hiding this comment.
can also remove the TODO comment
src/uct/api/device/uct_device_impl.h
Outdated
| #include <uct/ib/mlx5/gdaki/gdaki.cuh> | ||
|
|
||
| union uct_device_completion { | ||
| uct_rc_gda_completion_t rc; |
src/uct/ib/mlx5/gdaki/gdaki.cuh
Outdated
| 0xffffff); | ||
| uct_rc_mlx5_gda_qedump("WQE", wqe_ptr, 64); | ||
| uct_rc_mlx5_gda_qedump("CQE", cqe64, 64); | ||
| ep->wqe_error = ep->sq_wqe_pi; |
There was a problem hiding this comment.
Do we need some "atomic write" so it will be visible to all threads?
src/uct/ib/mlx5/gdaki/gdaki.cuh
Outdated
Comment on lines
600
to
604
| if (ep->wqe_error && comp->wqe_idx < ep->wqe_error) { | ||
| return UCS_ERR_IO_ERROR; | ||
| } | ||
|
|
||
| status = (ucs_status_t)__shfl_sync(0xffffffff, status, 0); | ||
| __syncwarp(); | ||
| return status; | ||
| } else if (level == UCS_DEVICE_LEVEL_THREAD) { | ||
| return uct_rc_mlx5_gda_progress_thread(ep); | ||
| } else { | ||
| return UCS_ERR_UNSUPPORTED; | ||
| if (ep->sq_wqe_pi <= comp->wqe_idx) { |
There was a problem hiding this comment.
- why need to check ep->wqe_error separately?
IMOcomp->wqe_idx < ep->wqe_erroris enough - do we need some kind of atomic read/write for ep->wqe_error?
- maybe rename ep->wqe_error to ep->wqe_error_pi?
- if another thread updated ep->sq_wqe_pi but didn't yet update ep->wqe_error, we might think some tl_comp is success, while in fact it is failure.
src/uct/ib/mlx5/gdaki/gdaki_dev.h
Outdated
| uint64_t cqe_ci; | ||
| int sq_lock; | ||
|
|
||
| uint64_t wqe_error; |
There was a problem hiding this comment.
maybe move it after sq_wqe_pi?
iyastreb
previously approved these changes
Oct 2, 2025
|
This PR was hanging in cuda kernel perftest. So may be there is a real issue. |
iyastreb
approved these changes
Oct 6, 2025
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.
No description provided.