aniqueakhtar
commented
3 years ago I have been able to get these two dockerfiles to work.
Maybe it would be better to include these dockerfiles in the documentation so others can follow it:
I added the TORCH_CUDA_ARCH_LIST="8.6+PTX" because it wasn't working without it for my GPU (RTX 3090).
Others would have to use TORCH_CUDA_ARCH_LIST="YOUR_GPUs_CC+PTX" to run:
More details:
https://pytorch.org/docs/stable/cpp_extension.html
Find your YOUR_GPUs_CC at:
https://developer.nvidia.com/cuda-gpus
This Dockerfile is working with 3090 Dockerfile # 1:
FROM nvidia/cuda:11.1.1-cudnn8-devel-ubuntu20.04
RUN apt-get update
RUN DEBIAN_FRONTEND=noninteractive apt-get install keyboard-configuration -y
RUN DEBIAN_FRONTEND=noninteractive apt-get install -y tzdata
RUN apt-get --assume-yes install software-properties-common
RUN add-apt-repository ppa:deadsnakes/ppa
#RUN apt install python3.7 -y
#RUN ln -s /usr/bin/python3.8 /usr/bin/python3
RUN ln -s /usr/bin/python3.8 /usr/bin/python
RUN apt install python3-pip -y
RUN python -m pip install --upgrade pip
#RUN python -m pip install --upgrade pip3
RUN apt-get install -y\
build-essential \
apt-utils \
ca-certificates \
wget \
git \
vim \
libssl-dev \
curl \
unzip \
unrar
RUN apt-get install -y libsm6 libxrender1 libfontconfig1 libpython3.7-dev libopenblas-dev
RUN apt install libgl1-mesa-glx -y
RUN python -m pip install torch==1.9.0+cu111 torchvision==0.10.0+cu111 torchaudio==0.9.0 -f https://download.pytorch.org/whl/torch_stable.html
RUN python -m pip install matplotlib \
&& python -m pip install pillow \
&& python -m pip install numpy \
&& python -m pip install scipy \
&& python -m pip install cython \
&& python -m pip install scikit-image \
&& python -m pip install sklearn \
&& python -m pip install opencv-python \
&& python -m pip install open3d
RUN TORCH_CUDA_ARCH_LIST="8.6+PTX" python -m pip install -U git+https://github.com/StanfordVL/MinkowskiEngine \
--install-option="--force_cuda" \
--install-option="--cuda_home=/usr/local/cuda" \
--install-option="--blas=openblas"
RUN python -m pip install h5py \
&& python -m pip install tensorboardX \
&& python -m pip install easydictThis Dockerfile I used with 1080Ti Dockerfile # 2:
FROM nvidia/cuda:10.2-cudnn7-devel-ubuntu18.04
RUN apt-get update
RUN DEBIAN_FRONTEND=noninteractive apt-get install keyboard-configuration -y
RUN apt install python3.7 -y
RUN ln -s /usr/bin/python3.7 /usr/bin/python3
RUN ln -s /usr/bin/python3.7 /usr/bin/python
RUN apt install python3-pip -y
RUN python -m pip install pip
RUN apt-get install -y\
build-essential \
apt-utils \
ca-certificates \
wget \
git \
vim \
libssl-dev \
curl \
unzip \
unrar
RUN apt-get install -y libsm6 libxrender1 libfontconfig1 libpython3.7-dev libopenblas-dev
RUN apt install libgl1-mesa-glx -y
RUN python -m pip install matplotlib \
&& python -m pip install pillow \
&& python -m pip install numpy \
&& python -m pip install scipy \
&& python -m pip install cython \
&& python -m pip install scikit-image \
&& python -m pip install sklearn \
&& python -m pip install opencv-python \
&& python -m pip install open3d \
&& python -m pip install torch \
&& python -m pip install torchvision
RUN python -m pip install -U git+https://github.com/StanfordVL/MinkowskiEngine \
--install-option="--force_cuda" \
--install-option="--cuda_home=/usr/local/cuda" \
--install-option="--blas=openblas"
houyongkuo
emilypython
haiduo
simenhu
previously, I was able to install Minkowski on 'RTX 1080 Ti' with 'gcc 7.5' using the following command:
But now using 'RTX 3090', I can't get it to run using the above command or any of these commands:
I also tried other variations.
Everything almost gets the same error:
Basically, I have tried combinations of Pytorch=1.7.1, 1.8, and 1.9, python 3.7 and 3.8, cudatoolkit=10.2, 11.0, and 11.1. I have tried both Anaconda and pip installation. Nothing seems to be working.
Searching around, it seems a lot of deep learning libraries are facing similar issues. Any help would be greatly appreciated. Thanks