colossalai

1
FROM hpcaitech/cuda-conda:11.3
2

3
# metainformation
4
LABEL org.opencontainers.image.source = "https://github.com/hpcaitech/ColossalAI"
5
LABEL org.opencontainers.image.licenses = "Apache License 2.0"
6
LABEL org.opencontainers.image.base.name = "docker.io/library/hpcaitech/cuda-conda:11.3"
7

8
# enable passwordless ssh
9
RUN mkdir ~/.ssh && \
10
    printf "Host * \n    ForwardAgent yes\nHost *\n    StrictHostKeyChecking no" > ~/.ssh/config && \
11
    ssh-keygen -t rsa -N "" -f ~/.ssh/id_rsa && \
12
    cat ~/.ssh/id_rsa.pub >> ~/.ssh/authorized_keys
13

14
# enable RDMA support
15
RUN apt-get update && \
16
    apt-get install -y infiniband-diags perftest ibverbs-providers libibumad3 libibverbs1 libnl-3-200 libnl-route-3-200 librdmacm1 && \
17
    apt-get clean && \
18
    rm -rf /var/lib/apt/lists/*
19

20
# install torch
21
RUN conda install -y pytorch==1.12.1 torchvision==0.13.1 torchaudio==0.12.1 cudatoolkit=11.3 -c pytorch
22

23
# install ninja
24
RUN apt-get update && \
25
    apt-get install -y --no-install-recommends ninja-build && \
26
    apt-get clean && \
27
    rm -rf /var/lib/apt/lists/*
28

29
# install apex
30
RUN git clone https://github.com/NVIDIA/apex && \
31
    cd apex && \
32
    git checkout 91fcaa && \
33
    pip install packaging && \
34
    pip install -v --disable-pip-version-check --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" --global-option="--fast_layer_norm" ./
35

36
# install colossalai
37
ARG VERSION=main
38
RUN git clone -b ${VERSION} https://github.com/hpcaitech/ColossalAI.git \
39
    && cd ./ColossalAI \
40
    && BUILD_EXT=1 pip install -v --no-cache-dir .
41

42
# install titans
43
RUN pip install --no-cache-dir titans
44

45
# install tensornvme
46
RUN conda install -y cmake && \
47
    git clone https://github.com/hpcaitech/TensorNVMe.git && \
48
    cd TensorNVMe && \
49
    apt update -y && apt install -y libaio-dev && \
50
    pip install -r requirements.txt && \
51
    pip install -v --no-cache-dir .
52