Adding SSE instructions to ggml_vec_dot_q4_0_q8_0 #1413
sw
merged
9 commits into
ggml-org:master
from
3ooabkhxtn:master
- added sse instructions for ggml_vec_dot_q4_0_q8_0
524d6c94
- Split multiplication and addition to make it easier for the compile…
607b9c73
- Use 4 accumulations instead of 2
78bbb3cd
- Added prefetch
7379dd2d
- Cleanup
8699fd0d
3ooabkhxtn
changed the title
- Improved prefetching
ca54314a
Put __SSE3__ into defined()
70c2b6c6
- Put the whole line into defined()
fc26f54e
github-actions
commented
on 2023-05-12
- rearranged defines, SSSE3 function only compiled if used
25b448a3
sw
approved these changes
on 2023-05-13
Login to write a write a comment.
Login via GitHub