Skip to content

Commit a59b4fc

Browse files
JohannesGaesslernjsyw1997
authored andcommitted
CUDA: faster tile FA (Pascal/AMD), headsize 256 (ggml-org#15769)
1 parent 7b23b9c commit a59b4fc

File tree

7 files changed

+604
-769
lines changed

7 files changed

+604
-769
lines changed

ggml/src/ggml-cuda/fattn-tile-f16.cu

Lines changed: 0 additions & 371 deletions
This file was deleted.

ggml/src/ggml-cuda/fattn-tile-f16.cuh

Lines changed: 0 additions & 3 deletions
This file was deleted.

0 commit comments

Comments
 (0)