You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Discussed internally. With all things considered will take a multi-phase approach to iteratively enhance the CUDA graph coverage. Below is the phase-1 design considerations:
Only cover stream capture (no explicit graph construction)
Exclude memory allocation/deallocation steps, and assume when entering the capturing context all needed memory are already allocated by the user
Exclude host callback operations
Basic coverage for conditional nodes
The resulting graph should be replay-able, meaning
Activity
leofang commentedon Nov 12, 2024
A few pointers to consider when we design this:
leofang commentedon Jan 23, 2025
Discussed internally. With all things considered will take a multi-phase approach to iteratively enhance the CUDA graph coverage. Below is the phase-1 design considerations:
[-]Design of graph support[/-][+]Design of graph support - Phase 1[/+]leofang commentedon Mar 24, 2025
Design is being wrapped up with a prototype (#455). Moving this to beta 4.