llama.cpp
Porting the improved K-Quant CUDA kernels to OpenCL
#1966

Merged

Porting the improved K-Quant CUDA kernels to OpenCL #1966

0cc4m merged 13 commits into ggml-org:master from LostRuins:optimize_quants_upstream

Added broken new q4k quant

266d4367

xx + ib0

c94a4383

Merge branch 'master' into optimize_quants_upstream

69fd31d1

Fix q2_k fast kernel

069cbe53

Use preprocessor for QK_K

34a49179

Add q6_k fast matmul kernel

8d816d19

ported q3k speedup successfully

029bed64

ported q2k and q5k speedups

93247a11

remove old dot kernels and template

a6e8b021

fixed global const struct types

6b75fc48

fixing address spaces

da668e68

Merge branch 'master' into optimize_quants_upstream

0eedccaf

LostRuins marked this pull request as ready for review 3 years ago

fixed string too long CI issue

f7b09637

howard0su commented on 2023-06-23

howard0su approved these changes on 2023-06-23

howard0su commented on 2023-06-23

ggerganov approved these changes on 2023-06-24

0cc4m merged 96a712ca into master 3 years ago

LostRuins deleted the optimize_quants_upstream branch 3 years ago

Reviewers

ggerganov

howard0su

0cc4m

Assignees

No one assigned

Labels

None yet

Milestone

No milestone