cdc_dockerfile/evo/Dockerfile

ARG CUDA_VERSION=12.1.0
FROM nvidia/cuda:${CUDA_VERSION}-cudnn8-devel-ubuntu20.04
ARG DEBIAN_FRONTEND="noninteractive"
ENV DEBIAN_FRONTEND=${DEBIAN_FRONTEND}
ENV MAMBA_ROOT_PREFIX=~/micromamba
ARG CONDA_ENV_NAME="deepspeed"
ENV CONDA_ENV_NAME=${CONDA_ENV_NAME}
ARG PYTHON_VERSION=3.10
ENV PYTHON_VERSION=${PYTHON_VERSION}
ARG ROOT_PASSWD="root"
ENV ROOT_PASSWD=${ROOT_PASSWD}
ENV PATH /opt/conda/bin:/opt/conda/envs/${CONDA_ENV_NAME}/bin:$PATH
WORKDIR /root
SHELL ["/bin/bash", "-c"]
# base tools
RUN <<EOT
#!/bin/bash
apt-get update
apt-get install -y wget curl htop jq vim bash libaio-dev build-essential openssh-server python3 python3-pip
apt-get install -y --no-install-recommends software-properties-common build-essential autotools-dev nfs-common pdsh cmake g++ gcc curl wget vim tmux emacs less unzip htop iftop iotop ca-certificates openssh-client openssh-server rsync iputils-ping net-tools sudo llvm-dev re2c
add-apt-repository ppa:git-core/ppa -y
apt-get install -y git libnuma-dev wget
# install latest cmake
wget -O - https://apt.kitware.com/keys/kitware-archive-latest.asc | sudo apt-key add -
sudo apt-add-repository "deb https://apt.kitware.com/ubuntu/ $(lsb_release -cs) main"
sudo apt-get update
sudo apt-get install -y cmake
# Configure SSH for password and public key authentication
sed -i 's/#PermitRootLogin prohibit-password/PermitRootLogin yes/' /etc/ssh/sshd_config
sed -i 's/#PasswordAuthentication yes/PasswordAuthentication yes/' /etc/ssh/sshd_config
sed -i 's/PubkeyAuthentication no/PubkeyAuthentication yes/' /etc/ssh/sshd_config
sed -i 's/^#Port 22/Port 22/' /etc/ssh/sshd_config
sed -i 's/^Port [0-9]*/Port 22/' /etc/ssh/sshd_config
mkdir /var/run/sshd
echo 'root:${ROOT_PASSWD}' | chpasswd
mkdir -p ~/.pip
# install miniconda
wget -qO- https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O /tmp/miniconda.sh
bash /tmp/miniconda.sh -b -p /opt/conda 
rm /tmp/miniconda.sh 
conda init bash 
ln -s /opt/conda/etc/profile.d/conda.sh /etc/profile.d/conda.sh
echo ". /opt/conda/etc/profile.d/conda.sh" >> ~/.bashrc 
echo "conda activate ${CONDA_ENV_NAME}" >> ~/.bashrc
# 配置 .condarc 文件
cat <<EOF > ~/.condarc
channels:
  - conda-forge
  - bioconda
  - pytorch
  - pytorch-nightly
  - nvidia
  - defaults
show_channel_urls: true
EOF
# 安装 micromamba
echo 1 | bash <(curl -s https://cdn.jsdelivr.net/gh/hotwa/MicroMamba_Installer@main/install.sh)
micromamba shell init -s bash -p ~/micromamba
cat <<'EOF' >> ~/.bashrc
source ~/micromamba/etc/profile.d/micromamba.sh
alias mamba=micromamba
alias mba=mamba
EOF
# 配置 .mambarc 文件
cat <<EOF > ~/.mambarc
channels:
  - conda-forge
  - bioconda
  - pytorch
  - pytorch-nightly
  - nvidia
  - defaults
show_channel_urls: true
EOF
EOT

# reference: https://github.com/huggingface/transformers/blob/main/docker/transformers-pytorch-deepspeed-latest-gpu/Dockerfile
# PyTorch
ENV REF='main'
ENV STAGE_DIR=/tmp
ENV NV_PEER_MEM_VERSION=1.2
ENV NV_PEER_MEM_TAG=${NV_PEER_MEM_VERSION}-0
ENV OPENMPI_BASEVERSION=4.1
ENV OPENMPI_VERSION=${OPENMPI_BASEVERSION}.6
ARG CUDA='cu121'
ENV CUDA=${CUDA}
ARG PYTORCH_VERSION=2.3.0
ENV PYTORCH_VERSION=${PYTORCH_VERSION}
ARG TORCHVISION_VERSION=0.18.0
ENV TORCHVISION_VERSION=${TORCHVISION_VERSION}
ARG TORCHAUDIO_VERSION=2.3.0
ENV TORCHAUDIO_VERSION=${TORCHAUDIO_VERSION}
ARG PYTORCH_CUDA_VERSION=12.1
ENV PYTORCH_CUDA_VERSION=${PYTORCH_CUDA_VERSION}
ENV MLNX_OFED_VERSION=4.9-7.1.0.0
ARG SETUPTOOLS_VERSION=69.5.1
ENV SETUPTOOLS_VERSION=${SETUPTOOLS_VERSION}
RUN <<EOT
#!/bin/bash
source /opt/conda/etc/profile.d/conda.sh
conda activate ${CONDA_ENV_NAME}
# 克隆 ninja 源码并编译
git clone https://github.com/ninja-build/ninja.git 
cd ninja
# 克隆 GoogleTest 源码
git clone https://github.com/google/googletest.git
conda run -n ${CONDA_ENV_NAME} python ./configure.py --bootstrap
# 配置并构建 Ninja 测试，添加 pthread 链接选项
# CXXFLAGS="-pthread" LDFLAGS="-pthread" ./configure.py --bootstrap --gtest-source-dir=$(pwd)/googletest
conda run -n ${CONDA_ENV_NAME} bash -c "CXXFLAGS='-pthread' LDFLAGS='-pthread' python ./configure.py --bootstrap --gtest-source-dir=$(pwd)/googletest"
./ninja all
# 运行 Ninja 单元测试
./ninja_test
conda create -n ${CONDA_ENV_NAME} python=${PYTHON_VERSION} pyyaml ipython -c conda-forge -y
python3 -m pip install --no-cache-dir --upgrade pip
conda clean -afy 
EOT

RUN <<EOT
#!/bin/bash
source /opt/conda/etc/profile.d/conda.sh
conda activate ${CONDA_ENV_NAME}
git clone https://github.com/evo-design/evo.git ${STAGE_DIR}/evo
cd ${STAGE_DIR}/evo
pip install packaging
python3 -m pip install torch==${PYTORCH_VERSION}+${CUDA} torchvision==${TORCHVISION_VERSION}+${CUDA} torchaudio==${TORCHAUDIO_VERSION} --extra-index-url https://download.pytorch.org/whl/${CUDA}
python3 -m pip install --no-cache-dir git+https://github.com/huggingface/accelerate@main#egg=accelerate
pip install .
EOT

RUN <<EOT
#!/bin/bash
source /opt/conda/etc/profile.d/conda.sh
conda activate ${CONDA_ENV_NAME}
pip install ipykernel
python3 -m ipykernel install --user --name=${CONDA_ENV_NAME} --display-name=${CONDA_ENV_NAME}
EOT

CMD ["/usr/sbin/sshd", "-D"]