llama.cpp
Make i-quants work with super-blocks of 64 (CPU and Metal)
#5760
Merged

Make i-quants work with super-blocks of 64 (CPU and Metal) #5760

ggerganov merged 6 commits into master from ik/i-quants-64
ikawrakow
WIP: make i-quants work for QK_K = 64
13ba37f1
iq2_xs: attempt to fix AVX dot product for QK_K = 64
28e6146c
QK_K = 64 tests pass on ARM_NEON and Metal
de64e061
Make CUDA compile with QK_K = 64
2540a290
Q2_K: fixed bug in imatrix quantization for QK_K = 64
47d52b2b
iq1_s: turn off SIMD implementation for QK_K = 64 (it does not work)
f0cbb6dd
ggerganov
ggerganov approved these changes on 2024-02-28
ggerganov ggerganov merged 7c4263d4 into master 1 year ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
Milestone