Repository navigation
Feature Request: CI/CD systems for llama.cpp+Vulkan #19213
Description
Prerequisites
- I am running the latest code. Mention the version if possible as well.
- I carefully followed the README.md.
- I searched using keywords relevant to my issue to make sure that I am creating a new issue that is not already open (or closed).
- I reviewed the Discussions, and have a new and useful enhancement to share.
Feature Description
There is lot of active development happening on llama.cpp+Vulkan. It seems like there is no CI/CD system for Intel platforms. We do have latest Intel platforms and can be used for this effort. How can we move forward?
Motivation
llama.cpp+Vulkan is very important for many developers. Great to see community is actively improving it. However, we are finding that many times there is regression or failure on Intel platforms. It would be great to catch them as early as possible and provide a stable release to the users.
Possible Implementation
We can provide access to Intel Hardware.
@ggerganov, @jeffbolznv and @0cc4m please let me know how we can proceed.
Activity
virajwad commentedon Jan 31, 2026
We initially saw vulkan perf regression when going from b7657 --> b7885 on one of our platforms
load_backend: loaded RPC backend from C:\Users\dungeon\Desktop\llamacpp\llama-b7657-bin-win-vulkan-x64\ggml-rpc.dll
ggml_vulkan: Found 1 Vulkan devices:
ggml_vulkan: 0 = Intel(R) Graphics (Intel Corporation) | uma: 1 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: KHR_coopmat
load_backend: loaded Vulkan backend from C:\Users\dungeon\Desktop\llamacpp\llama-b7657-bin-win-vulkan-x64\ggml-vulkan.dll
load_backend: loaded CPU backend from C:\Users\dungeon\Desktop\llamacpp\llama-b7657-bin-win-vulkan-x64\ggml-cpu-alderlake.dll
build: 0302329 (7657)
load_backend: loaded RPC backend from C:\Users\dungeon\Desktop\llamacpp\llama-b7885-bin-win-vulkan-x64\ggml-rpc.dll
ggml_vulkan: Found 2 Vulkan devices:
ggml_vulkan: 0 = Intel(R) Graphics (Intel Corporation) | uma: 1 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: KHR_coopmat
ggml_vulkan: 1 = Intel(R) Graphics (Intel Corporation) | uma: 1 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: KHR_coopmat
load_backend: loaded Vulkan backend from C:\Users\dungeon\Desktop\llamacpp\llama-b7885-bin-win-vulkan-x64\ggml-vulkan.dll
load_backend: loaded CPU backend from C:\Users\dungeon\Desktop\llamacpp\llama-b7885-bin-win-vulkan-x64\ggml-cpu-alderlake.dll
build: c3b87ce (7885)
After doing a painful and tedious binary search on the versions in between b7657 and b7885, I believe I narrowed it down to a regression going from b7858 --> b7860
I additionally filed this issue for it
#19221
It is great that llama.cpp has so many community contributions to support it, but it is difficult to keep up.
We'd love to help you guys get CI/CD for Intel platform in both functional and performance checks. We've also had things break in the past on Intel Vulkan llama.cpp side, so preventing early would be great.
ggerganov commentedon Jan 31, 2026
Setup a self-hosted runner with the hardware that you would like to CI (see https://docs.github.com/en/actions/concepts/runners/self-hosted-runners). Create github workflows and make sure they execute successfully with your runner - you can do this in a fork. When the runner is completely setup and the workflows are passing, let us know and we'll send you a token to add the runner to the organization. For reference, here is how we did this recently for RISC-V: #16682. Let me know if you have any questions.
bbharti commentedon Jan 31, 2026
Thanks @ggerganov. we will do.
github-actions commentedon Mar 17, 2026
This issue was closed because it has been inactive for 14 days since being marked as stale.