intel-analytics / ipex-llm

Accelerate local LLM inference and finetuning (LLaMA, Mistral, ChatGLM, Qwen, Mixtral, Gemma, Phi, MiniCPM, Qwen-VL, MiniCPM-V, etc.) on Intel XPU (e.g., local PC with iGPU and NPU, discrete GPU such as Arc, Flex and Max); seamlessly integrate with llama.cpp, Ollama, HuggingFace, LangChain, LlamaIndex, vLLM, GraphRAG, DeepSpeed, Axolotl, etc
Apache License 2.0
6.67k stars 1.26k forks source link

dGPU driver installation failed #11568

Open darshanhuang1 opened 3 months ago

darshanhuang1 commented 3 months ago

OS: 22.04.3 LTS, kernel default 6.5.0-41-general x86_64 When install dGPU driver following this guide (https://ipex-llm.readthedocs.io/en/latest/doc/LLM/Quickstart/install_linux_gpu.html), always encounter fail message as diagram showed. Even upgrade to 6.5.11, also failed.

MicrosoftTeams-image

liu-shaojun commented 3 months ago

Hi @darshanhuang1

We have verified this guide on kernel 6.5.0-35, could you try to use kernel 6.5.0-35? And I also tried 6.5.0-41, and I encountered problems when installing the driver.

darshanhuang1 commented 3 months ago

Ok. And how about 6.2 kernel? What versions have you verified?

liu-shaojun commented 3 months ago

We have verified 6.2.0-39-generic.

liu-shaojun commented 3 months ago

Hi @darshanhuang1,

We followed the instructions from this guide to install the GPU driver on kernel 6.5.0-41-generic. We found that intel-i915-dkms=1.24.2.17.240301.20+i29-1 encounters the same error, but upgrading to the latest version 1.24.3.23.240419.26+i30-1 resolves the issue.

If you are still using kernel 6.5.0-41-generic, you can execute the following commands to check if you can install the latest version:

sudo apt update
sudo apt-cache policy intel-i915-dkms

If you have the older version installed, remove it first and then install the latest one. If you cannot find the latest version in the candidate list, there might be an issue with the apt source.

sudo apt purge -y intel-i915-dkms
sudo apt install -y intel-i915-dkms 
darshanhuang1 commented 3 months ago

yes, work on 6.5.0-41-generic, will try anoter version soon. Thanks!

darshanhuang1 commented 3 months ago

Tried 6.6-intel kernel (Intel Tiber recommends kernel) but failed. Could the driver be supported on kernel 6.6?

liu-shaojun commented 3 months ago

We have currently only validated the driver on kernel versions 6.2 and 6.5. We have not yet tested it on kernel 6.6, so we are unable to confirm its compatibility with that version at this time.