[CUDA] Add opt-in SM121 L2 prefetch for quantized decode #33248
kunal-vaishnavi
wants to merge
5 commits into
main
from
redesigned-waffle
[CUDA] Add opt-in SM121 L2 prefetch for quantized decode
c750b76c
Merge remote-tracking branch 'origin/main' into fix-pr-33248
dc5856f4
[CUDA] Specify streams in quantized decode prefetch parity test
ecf55692
[CUDA] Specialize decode prefetch dispatch and validate on SM121
5c070efb
Merge main into redesigned-waffle and resolve NVFP4 dispatch conflict
6be8b8e9
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub