AI & Machine Learning11 min
FlashKDA: Moonshot's CUTLASS Kernel for Kimi Linear (2026)
Moonshot open-sourced FlashKDA, a CUTLASS CUDA kernel for Kimi Delta Attention. Drop-in for flash-linear-attention with up to 2.22x prefill speedup on H20 GPUs.
Tag
1 post
AI & Machine Learning11 min
Moonshot open-sourced FlashKDA, a CUTLASS CUDA kernel for Kimi Delta Attention. Drop-in for flash-linear-attention with up to 2.22x prefill speedup on H20 GPUs.