Kawrakow 52a7cbe482 Playing games with the scheduler ...

This change tricks it into doing the right thing^TM.
Still quite a bit slower than split mode layer for the 8B LlaMA model.
But for the 70B LlaMA it now beats split mode layer for TG:
28 t/s vs 24.4 t/s. PP is 627 t/s vs 744 t/s.
In comparison, split mode "row" in mainline gets
484 t/s PP and 19.3 t/s TG.

2025-11-30 18:05:13 +00:00

..

cmake

Merge mainline llama.cpp (#3)

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

include

WIP

2025-11-30 18:05:12 +00:00

src

Playing games with the scheduler

2025-11-30 18:05:13 +00:00

.gitignore

Merge mainline llama.cpp (#3)

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

CMakeLists.txt

Enable fusion by default (#939)

2025-11-11 10:35:48 +02:00