llama.cpp
metal : optimize ggml_mul_mat_id (faster Mixtral PP)
#4725
Merged
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Overview
Commits
17
Changes
View On
GitHub
metal : optimize ggml_mul_mat_id (faster Mixtral PP)
#4725
ggerganov
merged 17 commits into
master
from
gg/metal-opt-mul-mat-id
ggml : disable fast-math for Metal (cmake build only)
75c14f26
metal : fix Metal API debug warnings
515cfec4
cmake : add -fno-inline for Metal build (#4545)
a184e105
metal : fix API debug warnings
1580805f
metal : fix compile warnings
b14b5a9e
metal : use uint64_t for strides
4c054d98
cmake : rename option to LLAMA_METAL_SHADER_DEBUG
6435a3de
metal : fix mat-vec Q8_0 kernel for BS > 1
ad7cf37f
metal : normalize mat-vec kernel signatures
049a32ff
cmake : respect LLAMA_QKK_64 option
a8b9bb45
metal : fix mat-vec Q4_K kernel for QK_K == 64
5865b18e
metal : optimizing ggml_mul_mat_id (wip)
76f9d41d
Base automatically changed from
gg/fix-ci-metal
to
master
2 years ago
Merge branch 'master' into gg/metal-opt-mul-mat-id
c73e598d
Merge branch 'master' into gg/metal-opt-mul-mat-id
74460d00
metal : minor fix
daf9b124
Merge branch 'master' into gg/metal-opt-mul-mat-id
21e100d6
metal : opt mul_mm_id
9f51f3e6
ggerganov
merged
f3f62f0d
into master
2 years ago
Login to write a write a comment.
Login via GitHub
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub