diff options
| author | Kawrakow <iwankawrakow@gmail.com> | 2025-06-21 16:32:16 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2025-06-21 16:32:16 +0200 |
| commit | a98b7678a305c560117ce0a63a3529f2aaa17acb (patch) | |
| tree | 0e6e29d7cdbd7d1c335f6970032e9ea2f0dcf4cf /src/llama-vocab.cpp | |
| parent | 1843ed22c56cea6a4016005e78e26afd6c0c3948 (diff) | |
Perhaps slightly faster trellis quants (#541)
* This seems slightly faster for IQ2_KT, IQ3_KT TG
* This looks better for iq4_kt TG
* WIP
* Cleanup
* With fancy simd also set func16
* Enable next_128() also on AVX2
Despite having just 16 vector registers it is still faster.
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Diffstat (limited to 'src/llama-vocab.cpp')
0 files changed, 0 insertions, 0 deletions