Iwan Kawrakow 464b8fc03b CUDA: add head size of 64 to new mma ...

Haven't turned it on yet, but observe slightly better PP and slightly
worse TG performance with that.

2025-08-11 11:10:45 +03:00

..

cmake

Merge mainline llama.cpp (#3)

2024-07-27 07:55:01 +02:00

include

gpt-oss: attnetion sinks, swiglu_oai

2025-08-09 17:12:41 +03:00

src

CUDA: add head size of 64 to new mma

2025-08-11 11:10:45 +03:00

.gitignore

Merge mainline llama.cpp (#3)

2024-07-27 07:55:01 +02:00

CMakeLists.txt

Vulkan: add cmake options to build without coopmat(2) support (#674)

2025-08-07 17:26:21 +03:00