github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #4373] Feature request: support for OpenCL #64767

New Issue
Open
opened 2026-05-03 18:45:04 -05:00 by GiteaMirror · 7 comments
GiteaMirror commented 2026-05-03 18:45:04 -05:00
Owner
Copy Link

Originally created by @alnoses on GitHub (May 12, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/4373

Allow the possibility of running the models on those GPUs that do not support, have stopped supporting or have never had CUDA or ROCm support such as Intel (dedicated and integrated) or integrated AMD GPUs or old models such as Polaris, the offer Qualcomm ARM...

Originally created by @alnoses on GitHub (May 12, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/4373 Allow the possibility of running the models on those GPUs that do not support, have stopped supporting or have never had CUDA or ROCm support such as Intel (dedicated and integrated) or integrated AMD GPUs or old models such as Polaris, the offer Qualcomm ARM...
GiteaMirror added the feature request label 2026-05-03 18:45:04 -05:00
GiteaMirror commented 2026-05-03 18:45:06 -05:00
Author
Owner
Copy Link

@vt-alt commented on GitHub (Jul 7, 2024):

I wanted to suggest building with CLBlast support but it seems CLBlast support is recently removed from llama.cpp as "Superseded by Vulkan" https://github.com/ggerganov/llama.cpp/pull/7735

<!-- gh-comment-id:2212615845 --> @vt-alt commented on GitHub (Jul 7, 2024): I wanted to suggest building with CLBlast support but it seems CLBlast support is recently removed from `llama.cpp` as "Superseded by Vulkan" https://github.com/ggerganov/llama.cpp/pull/7735
GiteaMirror commented 2026-05-03 18:45:06 -05:00
Author
Owner
Copy Link

@serialscriptr commented on GitHub (Sep 8, 2024):

OpenCL support would provide better performance on SBCs such as the Orange Pi 5 or any other device that uses the rk3588. The Mali drivers have very capable opencl support but no vulken drivers.

<!-- gh-comment-id:2336861440 --> @serialscriptr commented on GitHub (Sep 8, 2024): OpenCL support would provide better performance on SBCs such as the Orange Pi 5 or any other device that uses the rk3588. The Mali drivers have very capable opencl support but no vulken drivers.
GiteaMirror commented 2026-05-03 18:45:07 -05:00
Author
Owner
Copy Link

@vpereira commented on GitHub (Feb 3, 2025):

with openCL we would be able to run small models (< 4 GB) straight from the GPUs.. it's interesting for instance for those running deepseek-r1:1.5b as local code completion tool..

<!-- gh-comment-id:2630685339 --> @vpereira commented on GitHub (Feb 3, 2025): with openCL we would be able to run small models (< 4 GB) straight from the GPUs.. it's interesting for instance for those running `deepseek-r1:1.5b` as local code completion tool..
GiteaMirror commented 2026-05-03 18:45:08 -05:00
Author
Owner
Copy Link

@pjaaskel commented on GitHub (Feb 3, 2025):

https://www.qualcomm.com/developer/blog/2024/11/introducing-new-opn-cl-gpu-backend-llama-cpp-for-qualcomm-adreno-gpu

<!-- gh-comment-id:2630733675 --> @pjaaskel commented on GitHub (Feb 3, 2025): https://www.qualcomm.com/developer/blog/2024/11/introducing-new-opn-cl-gpu-backend-llama-cpp-for-qualcomm-adreno-gpu
GiteaMirror commented 2026-05-03 18:45:08 -05:00
Author
Owner
Copy Link

@fish4terrisa-MSDSM commented on GitHub (Apr 7, 2025):

Any updates? Seems that OpenCL support is brought back to llama.cpp and is no longer tied with clblast now.

<!-- gh-comment-id:2782004347 --> @fish4terrisa-MSDSM commented on GitHub (Apr 7, 2025): Any updates? Seems that OpenCL support is brought back to llama.cpp and is no longer tied with clblast now.
GiteaMirror commented 2026-05-03 18:45:09 -05:00
Author
Owner
Copy Link

@dan-and commented on GitHub (May 6, 2025):

Also PoCL (https://github.com/pocl/pocl/blob/release_7_0/doc/sphinx/source/notes_7_0.rst ) would allow a single interface to a heterogenic set of backend devices

<!-- gh-comment-id:2854424333 --> @dan-and commented on GitHub (May 6, 2025): Also PoCL (https://github.com/pocl/pocl/blob/release_7_0/doc/sphinx/source/notes_7_0.rst ) would allow a single interface to a heterogenic set of backend devices
GiteaMirror commented 2026-05-03 18:45:09 -05:00
Author
Owner
Copy Link

@vpereira commented on GitHub (May 6, 2025):

Also PoCL (https://github.com/pocl/pocl/blob/release_7_0/doc/sphinx/source/notes_7_0.rst ) would allow a single interface to a heterogenic set of backend devices

it however doesn't support AMD GPUs, AFAICS...

<!-- gh-comment-id:2854862371 --> @vpereira commented on GitHub (May 6, 2025): > Also PoCL (https://github.com/pocl/pocl/blob/release_7_0/doc/sphinx/source/notes_7_0.rst ) would allow a single interface to a heterogenic set of backend devices it however doesn't support AMD GPUs, AFAICS...
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label feature request
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#64767
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?