Allow use of OpenCL GPU-based BLAS using ClBlast instead of OpenBLAS …
a908c37c
Improve ClBlast implementation, avoid recreating buffers, remove redu…
b7143c1a
Finish merge of ClBlast support
6f668707
Move CLBlast implementation to separate file
1b16b8c9
Add q4_2 and q4_3 CLBlast support, improve code
309af7fc
Double CLBlast speed by disabling OpenBLAS thread workaround
f469d9af
Fix device selection env variable names
8603c25e
Fix cast in opencl kernels
18cc05bd
Add CLBlast to CMakeLists.txt
ae73887f
Replace buffer pool with static buffers a, b, qb, c
daa5df51
0cc4m
changed the title Clblast llama cpp CLBlast support 2 years ago
slaren
commented
on 2023-04-25
slaren
commented
on 2023-04-25
slaren
commented
on 2023-04-25
Fix typos, use GGML_TYPE defines, improve code
36bfb3c1
slaren
commented
on 2023-04-25
SlyEcho
approved these changes
on 2023-04-25
Improve btype dequant kernel selection code, add error if type is uns…
13707100
slaren
approved these changes
on 2023-04-25
Improve code quality
2b0c6a56
Use c compiler for opencl files
b7464582
Simplify code, fix include
ce97a807
First check error, then release event
4a35ec9d
Make globals static, fix indentation
fafebff5
Rename dequant kernels file to conform with other file names
96346fb2
Fix import cl file name
bbfba5f7
SlyEcho
approved these changes
on 2023-04-27
ggerganov
approved these changes
on 2023-04-28
Merge branch 'master' into clblast-llama-cpp
4530d5c3
ggerganov
merged
7296c961
into master 2 years ago
0cc4m
deleted the clblast-llama-cpp branch 2 years ago
Assignees
No one assigned
Login to write a write a comment.
Login via GitHub