CUDA: refactor mmq, dmmv, mmvq (#7716)

Browse Source

* CUDA: refactor mmq, dmmv, mmvq

* fix out-of-bounds write

* struct for qk, qr, qi

* fix cmake build

* mmq_type_traits

...

This commit is contained in:

Johannes Gäßler

2024-06-05 16:53:00 +02:00

committed by GitHub

parent 2b3389677a

commit 7d1a378b8f

112 changed files with 1783 additions and 1767 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

1300

ggml-cuda/mmq.cuh

File diff suppressed because it is too large Load Diff