# # Copyright (c) 2025 Baidu Technologies Co., Ltd. All Rights Reserved. # This file is a part of the vllm-kunlun project. # # This file is mainly Adapted from vllm-project/vllm/vllm/envs.py # Copyright 2023 The vLLM team. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # FROM iregistry.baidu-int.com/xmlir/xmlir_ubuntu_2004_x86_64:v0.37_base SHELL ["bash", "-c"] ARG DEBIAN_FRONTEND=noninteractive RUN apt-get update && \ apt-get install -y --no-install-recommends \ ca-certificates tzdata vim net-tools \ gcc g++ cmake libnuma-dev \ wget tmux curl \ software-properties-common && \ apt-get clean && \ rm -rf /var/lib/apt/lists/* RUN conda init && conda create --name vllm_kunlun_0.10.1.1 python=3.10.15 -y && \ conda run -n vllm_kunlun_0.10.1.1 && source activate vllm_kunlun_0.10.1.1 && \ conda clean -afy RUN source activate vllm_kunlun_0.10.1.1 && \ pip install torch==2.5.1+cu118 torchvision==0.20.1+cu118 torchaudio==2.5.1+cu118 --index-url https://download.pytorch.org/whl/cu118 && \ pip cache purge && rm -rf /root/.cache/pip RUN source activate vllm_kunlun_0.10.1.1 && \ pip install setuptools==80.9.0 cuda_mock==1.1.1 hyperparameter==0.5.6 black==23.3.0 lark==1.2.2 \ networkx wheel PyYAML==6.0.2 ipython h5py regex==2024.9.11 colorama==0.4.6 pynvml==11.5.3 \ nvidia-cuda-runtime-cu11 tabulate==0.9.0 openpyxl==3.1.5 pandas prettytable \ pytest==8.1.0 pytest-repeat==0.9.3 pytest-timeout==2.3.1 py==1.11.0 datasets==2.16.0 \ pydantic==2.9.2 psutil==6.1.0 einops==0.8.0 \ pytest-html==4.1.1 py-cpuinfo pytest-timeout==2.3.1 termcolor jsonlines==4.0.0 tiktoken \ qwen_vl_utils filetype fastapi==0.112.1 fire py-cpuinfo transformers==4.57.1 \ gradio==4.0.0 sse-starlette trl==0.8.6 uvicorn accelerate==0.30.1 --index-url https://pip.baidu-int.com/simple/ && \ pip cache purge && rm -rf /root/.cache/pip ENV DEBIAN_FRONTEND=noninteractive \ LANG=en_US.UTF-8 \ LANGUAGE=en_US:en \ TERM=xterm-256color \ PATH=/root/miniconda/envs/vllm_kunlun_0.10.1.1/bin/:$PATH WORKDIR /workspace RUN wget https://su.bcebos.com/v1/klx-sdk-release-public/xccl/resource/MLNX_OFED_LINUX-24.01-0.3.3.1-ubuntu20.04-x86_64.tgz && tar -xf MLNX_OFED_LINUX-24.01-0.3.3.1-ubuntu20.04-x86_64.tgz WORKDIR /workspace/MLNX_OFED_LINUX-24.01-0.3.3.1-ubuntu20.04-x86_64 RUN apt-get install -y -o Dpkg::Options::='--force-confdef' -o Dpkg::Options::='--force-confold' flex swig tk debhelper libltdl-dev libusb-1.0-0 tcl chrpath pkg-config graphviz bison && \ ./mlnxofedinstall --user-space-only --skip-distro-check --without-fw-update --force && \ rm -rf /workspace/MLNX_OFED_LINUX-24.01-0.3.3.1-ubuntu20.04-x86_64* && \ apt-get clean && \ rm -rf /var/lib/apt/lists/* WORKDIR /workspace COPY vllm-kunlun/ /workspace/vllm-kunlun/ RUN cd /workspace/vllm-kunlun && \ bash dockerfile/install.sh && \ rm -rf /workspace/vllm-kunlun/build/ && rm -rf /workspace/vllm-kunlun/vllm_kunlun.egg-info/ # xpu-smi tools WORKDIR /workspace COPY xre-Linux-x86_64-5.2.0.0/ /workspace/xre-Linux-x86_64-5.2.0.0/ RUN mv /workspace/xre-Linux-x86_64-5.2.0.0/bin/* /usr/local/bin/ && mv /workspace/xre-Linux-x86_64-5.2.0.0/so/* /lib/x86_64-linux-gnu/ && \ rm -rf /workspace/xre-Linux-x86_64-5.2.0.0/ RUN rm -rf \ /root/.cache \ /root/.conda \ /tmp/* CMD ["/bin/bash"]