Skip to content

feat: fused turbo MMA flash-attention decode (turbo4/3/2, GQA tensor-core path) - #4

Merged
thecodacus merged 4 commits into
perffrom
fable5/turbo-mma-decode
Jul 18, 2026
Merged

feat: fused turbo MMA flash-attention decode (turbo4/3/2, GQA tensor-core path)#4
thecodacus merged 4 commits into
perffrom
fable5/turbo-mma-decode

fix(fattn): gate turbo2 fused-MMA decode to head_dim 128

e2f5f2f
Select commit
Loading
Failed to load commit list.

Select a check to view from the sidebar