Replace CUPTI callback with delayed init for NVIDIA GPUs#1250
Open
sanrise wants to merge 1 commit into
Open
Conversation
Summary: Previously, kineto registered a permanent CUPTI callback on CUDA context creation to initialize the profiler. This blocked external tools like wprof from subscribing to CUPTI since only one subscriber is allowed. This change replaces the CUPTI callback mechanism with a delayed initialization approach using a condition_variable-based timer. The profiler now initializes 1 second after libkineto_init, freeing CUPTI for on-demand profiling tools. The new DelayedInitializer class supports immediate cancellation on shutdown, avoiding the blocking behavior of std::future destructors. Differential Revision: D92570679
sanrise
force-pushed
the
export-D92570679
branch
from
February 12, 2026 00:38
c5d7067 to
3a2d638
Compare
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary:
Previously, kineto registered a permanent CUPTI callback on CUDA context creation to initialize the profiler. This blocked external tools like wprof from subscribing to CUPTI since only one subscriber is allowed.
This change replaces the CUPTI callback mechanism with a delayed initialization approach using a condition_variable-based timer. The profiler now initializes 1 second after libkineto_init, freeing CUPTI for on-demand profiling tools. The new DelayedInitializer class supports immediate cancellation on shutdown, avoiding the blocking behavior of std::future destructors.
Differential Revision: D92570679