ik_llama_opt/ggml
Kawrakow 0449090ae8 Zen4 Flash Attnetion 2 (#36)
* Zen4 Flash Attnetion: WIP generalize to other types

Now loading of data from K and V is done via a template parameter,
so this should make it easy to generalize to typ[es other than
F16 for the K and V cache.

* Zen4 Flash Attnetion: it works for q4_0 and q8_0

* Zen4 Flash Attnetion: small q8_0 performance improvement

* Zen4 Flash Attnetion: add q4_1

* Delete unused stuff

---------

Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2024-09-04 07:20:55 +03:00
..
cmake Merge mainline llama.cpp (#3) 2024-07-27 07:55:01 +02:00
include Faster Gemma2 (#27) 2024-08-27 17:40:59 +03:00
src Zen4 Flash Attnetion 2 (#36) 2024-09-04 07:20:55 +03:00
.gitignore Merge mainline llama.cpp (#3) 2024-07-27 07:55:01 +02:00
CMakeLists.txt Merge mainline - Aug 12 2024 (#17) 2024-08-12 15:14:32 +02:00