Add check and agent guidance about uncapped mempools - #2514
Open
juenglin wants to merge 2 commits into
Open
Conversation
Contributor
|
Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually. Contributors can view more details about this message here. |
juenglin
added
P1
Contributor
Author
|
/ok to test |
|
Andy-Jost
reviewed
Aug 5, 2026
Review feedback on NVIDIA#2514: - Move the uncapped-pool check out of the live test suite into a check-mempool-hygiene pre-commit hook. The rule is about source text and needs no GPU, so a hook catches it earlier and for free. Its tests move to ci/tools/tests, alongside the other check scripts'. - Add helpers/constants.py and route the eleven ad-hoc POOL_SIZE definitions through it. - Qualify "device memory" as installed/physical where the doc explains what an uncapped pool reserves.
Contributor
Author
|
/ok to test |
juenglin
marked this pull request as ready for review
August 5, 2026 23:43
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Follow-up to #2484, which right-sized seven memory pools that reserved an
address-space window scaling with device memory while holding at most 1 KiB.
This PR adds the guidance and the automated check to keep new pools from
reintroducing the problem.
cuda_core/tests/AGENTS.md(new) documents the rule: a pool created withoutmax_sizereserves roughly 1x device memory of address space regardless ofwhat the test allocates, and the whole suite shares one process, so those
reservations accumulate. It also spells out a distinction that is easy to get
backwards:
DeviceMemoryResource(dev)with no options does not create apool — it wraps the device's default mempool and costs nothing — so adding
max_sizeto such a call converts a free wrapper into a new pool and makesthings worse. Managed pools are documented as exempt, since
cuMemPoolCreaterequires
maxSize == 0for them.A new pre-commit check
check_mempool_hygieneenforces it. It statically parsesevery
.pyundertests/and fails onDeviceMemoryResourceOptions/PinnedMemoryResourceOptionsconstructions that omitmax_size, including thedict-literal options form. The check is static rather than runtime so it also
covers pools in tests that skip on the current platform. Deliberate exceptions
opt out with a
# uncapped-pool-ok: <reason>comment.Applying the check surfaced three call sites, all in
test_memory.pyand allinside
pytest.raiseswhere validation rejects the arguments before a pool iscreated; those are annotated.
test_multiprocessing_warning.pywas hardcoding2097152in four places and now uses the suite-standardPOOL_SIZEconstant.No production code changes; test-suite hygiene only.
Checklist