Skip to content

Conversation

sw
Copy link
Contributor

@sw sw commented Apr 22, 2023

#1109 was not finished for AVX (note: that affects all quantized formats, not just Q4_3 as the summary would suggest). This fixes it by introducing hsum_i32_4, in order to calculate s0 and s1.

@sw sw closed this Apr 22, 2023
@sw sw deleted the q8-avx branch April 22, 2023 08:11
@ggerganov
Copy link
Member

I added commented flags to the Makefile that can be used to go in AVX-only mode for easier debugging in the future:

https://github.com/ggerganov/llama.cpp/blob/master/Makefile#L79-L83

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants