Skip to content

Conversation

@pvelesko
Copy link
Collaborator

@pvelesko pvelesko commented Dec 20, 2025

@pvelesko pvelesko force-pushed the fix-11260-launchhost branch from c84bf6f to 3098147 Compare January 12, 2026 15:18
- Adds explicit host synchronization on queue sync events before adding barriers that wait on them.
- Introduced a copy command list barrier to synchronize pending copy operations with compute operations.
- Added a dedicated copy marker event to ensure proper synchronization between command lists.
- Updated the barrier logic to wait on both queue sync events and the new copy marker, improving event visibility and reducing potential race conditions.
- Add synchronization on both compute and copy command lists before callback
- Submit GPU barriers after callback completes to avoid race with host-signaled events
- Signal GpuAck and CallbackComplete from host for reliable completion
- Add CallbackComplete member to base CallbackData class
- Force command list flush after barrier submission
@pvelesko pvelesko force-pushed the fix-11260-launchhost branch from 3098147 to b7c539b Compare January 13, 2026 17:16
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

hipLaunchHostFunc Hangs Intermittently on Aurora with Level Zero Backend

2 participants