UCT/CUDA: Detect sys_dev for async allocations#10607
Merged
yosefe merged 1 commit intoopenucx:masterfrom Apr 7, 2025
Merged
Conversation
brminich
changed the title
Currently, asynchronous memory can be detected as CUDA-managed, with its sys_dev set to unknown. However, for async VMM memory, knowing the sys_dev is crucial to identify the correct CUDA context for executing cuMemcpyAsync.
brminich
force-pushed
the
uct/fix_async_mem_detection
branch
from
d8ebb65 to
88de9de
Compare
rakhmets
approved these changes
Apr 4, 2025
yosefe
approved these changes
Apr 7, 2025
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
3 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What?
Currently, asynchronous memory can be detected as CUDA-managed, with its sys_dev set to unknown. However, for async VMM memory, knowing the sys_dev is crucial to identify the correct CUDA context for executing cuMemcpyAsync.
Why?
Test from #10601 fails when sending an eager message from legacy pinned memory to asynchronous VMM memory (on isr1). On the receiver side, the destination buffer is detected as CUDA-managed with an unknown sys_device, preventing the cuda_copy transport from selecting the correct context for the VMM allocation.