Retry test failures - #617
Merged
Merged
Conversation
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #617 +/- ##
=======================================
Coverage 80.48% 80.48%
=======================================
Files 50 50
Lines 3490 3490
=======================================
Hits 2809 2809
Misses 681 681 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
Member
|
ALCF pipeline is broken after a maintenance window. Filed a ticket. |
michel2323
added
alcf-ci
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
michel2323
added
alcf-ci
michel2323
referenced
this pull request
Aug 21, 2026
Every kernel launch, copy, and fill used to create a fresh command list, submit it to the task's command queue, and drop the reference — leaving destruction of thousands of driver objects (lists, command buffers, heaps) to finalizer timing. Under launch storms that garbage is what pushes the driver into allocation failure, where NEO's error handling is at its worst (the scratch path aborts outright). It is also pure overhead: the per-dispatch list costs ~8x in submission latency. Replace the per-dispatch machinery with a per-task oneStream holding one in-order asynchronous immediate command list: appends submit directly, and the garbage source disappears entirely. Level Zero >= 1.9 is required; there is no fallback submission path. oneMKL work still needs a real command queue for SYCL interop, so each stream lazily creates a companion queue — a separate execution stream, which makes the previously implicit ordering between Julia kernels and oneMKL calls explicit: sycl_queue drains the immediate list before handing out the SYCL queue (Julia -> MKL), and a dirty flag makes the next Julia-side submission drain the companion queue (MKL -> Julia). FFT plans capture their queue at construction, so their _exec! methods apply the boundary themselves. The LTS drain-before-free machinery follows the shape change: the queue registry becomes a stream registry, draining both the immediate list and the companion queue before a buffer referenced by in-flight work is freed; immediate lists get the same bounded-drain finalizer as queues. The sync-each-submission workaround now host-synchronizes the list after each append. KA.priority! swaps the task's stream for one with the requested priority. The scratch hedge moves to the stream, and remains on the explicit-queue compatibility path (@oneapi queue=...), which still submits through a per-dispatch list.
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.
Revert the hacky workaround.
Runic fails because this branch is in a fork but it was run