Add regression coverage for CUDA-context-independent topology discovery - #191
Merged
rapids-bot[bot] merged 2 commits intoSep 16, 2026
Merged
Conversation
Topology discovery must be safe to use independently of CUDA runtime and context state. It should not initialize CUDA, require an active context, create a context, or otherwise change CUDA process state. This change restores that invariant that was previously modified in NVIDIA#167 and documents it as part of the topology-discovery API contract. It also adds GPU coverage to guard the behavior going forward.
pentschev
force-pushed
the
fix-topology-cuda-context-agnostic
branch
from
August 18, 2026 14:58
b5bb37a to
7aff354
Compare
dhruv9vats
approved these changes
Aug 18, 2026
aminaramoon
approved these changes
Aug 18, 2026
felipeblazing
requested changes
Aug 18, 2026
| return false; | ||
| } | ||
| constexpr int min_hw_decompression_cuda_version = 12080; | ||
| int driver_version = 0; |
Contributor
There was a problem hiding this comment.
not all hardware has a DE engine even if the driver supports it
Contributor
Author
There was a problem hiding this comment.
Yes, you're right. I realized that and have been discussing this offline as well, I'll link to the existing discussion, in the meantime we should hold merging this.
This was referenced Aug 26, 2026
…ontext-agnostic # Conflicts: # include/cucascade/memory/topology_discovery.hpp # src/memory/topology_discovery.cpp
pentschev
changed the title
felipeblazing
approved these changes
Sep 16, 2026
Contributor
|
/merge |
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
4 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
PR #176 keeps default topology discovery independent of CUDA runtime and context state. This PR closes the remaining coverage gap by adding a multi-GPU regression test that verifies default topology discovery does not initialize CUDA or capture temporarily widened
CUDA_VISIBLE_DEVICES, as well as documenting that default discovery is CUDA-context independent, while runtime-attribute discovery is explicitly opt-in.