CUDA: optimize and refactor MMQ (#8416)

Browse Source

* CUDA: optimize and refactor MMQ

* explicit q8_1 memory layouts, add documentation

...

This commit is contained in:

Johannes Gäßler

2024-07-11 16:47:47 +02:00

committed by GitHub

parent a977c11544

commit 808aba3916

5 changed files with 867 additions and 687 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download Patch File Download Diff File Expand all files Collapse all files

ggml/src/ggml-cuda/mmq.cuh

+727 -638

File diff suppressed because it is too large Load Diff