Environment Requirements
This document uses openEuler 24.03 LTS SP3 + vLLM 0.18.1 + CUDA 13.0 as an example to describe how to verify the installation of the GPU path. In addition, this section describes other environments supported by vLLM installed on the Kunpeng server.
Supported Versions
OS |
Python |
vLLM |
|---|---|---|
openEuler 24.03 LTS SP3 |
Python 3.11.6 |
vLLM 0.18.1 |
openEuler 22.03 LTS SP4 |
Python 3.9.9 |
vLLM <= 0.11.x (vLLM 0.11.0 is recommended.) |
vLLM 0.18.x is designed for newer Python versions. The default Python version of openEuler 22.03 LTS SP4 is 3.9.9, which is not suitable as the default build environment for vLLM 0.18.x. You are advised to use vLLM 0.11.0 that officially supports Python 3.9.
Dependent Component Requirements
Component |
Version |
|---|---|
python3 |
3.11.6-20.oe2403sp3 |
python3-pip |
23.3.1-7.oe2403sp3 |
python3-devel |
3.11.6-20.oe2403sp3 |
pip |
26.1.1 |
PyTorch |
2.11.0 CUDA version |
CUDA Toolkit |
13.0 |
vLLM |
0.18.1 |
To use the GPU capability, the system must provide an accessible NVIDIA GPU device, CUDA Toolkit, and the matching NVIDIA driver user-mode library.