llama.cpp
Introducing experimental OpenCL backend with support for Qualcomm Adreno GPUs
#10693
Merged
Overview
Commits
28
Changes
View On GitHub

Introducing experimental OpenCL backend with support for Qualcomm Adreno GPUs #10693

max-krasnyansky merged 28 commits into ggml-org:master from CodeLinaro:adreno-support
lhez
github-actions github-actions added python
github-actions github-actions added devops
github-actions github-actions added ggml
max-krasnyansky max-krasnyansky requested a review from slaren slaren 1 year ago
max-krasnyansky max-krasnyansky removed review request from slaren slaren 1 year ago
max-krasnyansky max-krasnyansky requested a review from ggerganov ggerganov 1 year ago
max-krasnyansky
max-krasnyansky max-krasnyansky requested a review from slaren slaren 1 year ago
ggerganov
ggerganov commented on 2024-12-07
oscarbg
netrunnereve
myan-o
max-krasnyansky
max-krasnyansky
max-krasnyansky
myan-o
slaren
slaren commented on 2024-12-10
myan-o
max-krasnyansky
myan-o
lhez
myan-o
lhez lhez force pushed 1 year ago
slaren
slaren commented on 2024-12-11
netrunnereve
max-krasnyansky
max-krasnyansky
max-krasnyansky
max-krasnyansky
max-krasnyansky
max-krasnyansky approved these changes on 2024-12-12
slaren
max-krasnyansky
slaren
max-krasnyansky
slaren
max-krasnyansky
lhez [cl][adreno] Add Adreno GPU support
f56fb699
lhez [cl][ci] Add workflow for CL
3571bb6c
lhez [cl][adreno] Fix memory leak for non SMALL_ALLOC path
c1af4b72
max-krasnyansky opencl: integrate backend dyn.load interface and fix compiler and for…
8ad0bb30
max-krasnyansky opencl: remove small-alloc support and fix build errors for non-openc…
671c7af6
max-krasnyansky opencl: fixed merge conflict (MUSA added twice in cmake)
d24b3602
max-krasnyansky opencl-ci: use RUNNER_TEMP instead of github.workspace
9b6540b6
max-krasnyansky opencl: fix embed tool invocation with python3
4bca601b
max-krasnyansky opencl: CI workflow fixes
969a00a4
lhez opencl: Clean up small-alloc in CMake files
66d43303
max-krasnyansky opencl: cleanup ggml-opencl2 header file
0451edd9
max-krasnyansky opencl: use ulong for offsets and strides in ADD kernel
31f305ea
max-krasnyansky opencl: use cl_ulong for all offsets
c21fc8c5
max-krasnyansky opencl: use cl_ulong for sizes and strides
9a9d92b0
lhez opencl: use `GGML_LOG_xxx` instead of `fprintf(stderr, ...)`
e9a97381
lhez opencl: rename backend `opencl2` -> `opencl`
34f2fc15
lhez opencl: rename kernel files `ggml-opencl2` -> `ggml-opencl`
97a12703
lhez opencl: make OpenCL required, remove redundant lib and inc directories
22411ab5
lhez opencl: rename backend - funcs, structs, etc `opencl2` -> `opencl`
e447dbcc
lhez opencl: remove copyright marker since main license already covers
c64ef0fb
max-krasnyansky opencl: replace some more OPENCL2 leftovers
70063c6c
lhez opencl: remove limits on `tensor_extra`
74a9bafc
lhez opencl: use pools for `tensor_extra`
3bc085b3
max-krasnyansky opencl: fix compiler warnings with GCC and Clang
c971a188
max-krasnyansky opencl: fail gracefully if opencl devices are not available
b25a4caa
max-krasnyansky opencl: fix MSVC builds (string length error)
b41b6e67
lhez opencl: check for various requirements, allow deprecated API
dbaa360a
max-krasnyansky opencl: update log message for unsupported GPUs
9697d07b
max-krasnyansky max-krasnyansky force pushed to 9697d07b 1 year ago
max-krasnyansky
slaren
slaren approved these changes on 2024-12-13
max-krasnyansky max-krasnyansky merged a76c56fa into master 1 year ago
AndreasKunar
max-krasnyansky
AndreasKunar
sherylynn
sherylynn
myan-o
sherylynn
max-krasnyansky
max-krasnyansky
AndreasKunar
slaren
sherylynn
myan-o
Francis235

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
python devops ggml
Milestone
No milestone
Terms Of Use Privacy Policy Refund Policy Impressum