Skip to content

cuda: PTQ1_0 decode on consumer Ampere/Ada: SoA q8 activations, warp-per-row small-K GEMV (+10.5% TG, +14.7% with #220) - #215

Closed
professorpalmer wants to merge 3 commits into
PrismML-Eng:prismfrom
professorpalmer:cuda-ptq1-decode
Closed

cuda: PTQ1_0 decode on consumer Ampere/Ada: SoA q8 activations, warp-per-row small-K GEMV (+10.5% TG, +14.7% with #220)#215
professorpalmer wants to merge 3 commits into
PrismML-Eng:prismfrom
professorpalmer:cuda-ptq1-decode

cuda: restore the scalar PTQ1_0 vec-dot on MUSA

68ea9b4
Select commit
Loading
Failed to load commit list.
Sign in for the full log view

Annotations

1 warning
labeler
succeeded Sep 25, 2026 in 20s