Skip to content

[CUDA] Use Triton TMA for long-context causal prefill - #22414

Open
Gasoonjia wants to merge 2 commits into
codex/stack-d113-persistent-splitkfrom
codex/stack-tma-prefill
Open

[CUDA] Use Triton TMA for long-context causal prefill#22414
Gasoonjia wants to merge 2 commits into
codex/stack-d113-persistent-splitkfrom
codex/stack-tma-prefill

cuda: gate TMA causal prefill behind compile spec

8da3f29
Select commit
Loading
Failed to load commit list.

Select a check to view from the sidebar