]> git.djapps.eu Git - pkg/ggml/sources/ggml/commit
CUDA: skip fully masked-out KV in FA vec kernel (llama/13584)
authorJohannes Gäßler <redacted>
Tue, 20 May 2025 12:45:07 +0000 (14:45 +0200)
committerGeorgi Gerganov <redacted>
Sun, 25 May 2025 07:46:24 +0000 (10:46 +0300)
commit6b131860f3df34dc3637b633525482ac03b326bd
tree638bfc620624de798653d4d6b2288d08df8f3390
parent709ec018eec7cb8a45c171eab754a442b8baccfa
CUDA: skip fully masked-out KV in FA vec kernel (llama/13584)

* CUDA: skip fully masked-out KV in FA vec kernel
src/ggml-cuda/fattn-vec-f16.cuh
src/ggml-cuda/fattn-vec-f32.cuh