Added torch-gpu Dockerfile in packaging/dockers (#2531)

* Added torch-gpu Dockerfile in packaging/dockers

Signed-off-by: Manvenddra Rawat <quic_manvendd@quicinc.com>

packaging/dockers/Dockerfile.torch_gpu

@@ -0,0 +1,74 @@
+FROM docker-registry.qualcomm.com/library/nvidia/cuda:11.6.2-cudnn8-devel-ubuntu20.04
+
+ARG DEBIAN_FRONTEND=noninteractive
+ARG APT_KEY_DONT_WARN_ON_DANGEROUS_USAGE=DontWarn
+
+RUN mv /etc/apt/sources.list.d/cuda.list /etc/apt/sources.list.d/cuda.list.orig && \
+    apt-get update > /dev/null && \
+    apt-get install -y --no-install-recommends apt-utils && \
+    apt-key del --no-tty 7fa2af80 && \
+    apt-key adv --no-tty --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/3bf863cc.pub && \
+    apt-key adv --no-tty --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/7fa2af80.pub && \
+    apt-get update > /dev/null && \
+    rm -rf /var/lib/apt/lists/*
+
+RUN apt-get update > /dev/null && \
+    apt-get install --no-install-recommends -y \
+        # Bare minimum Packages
+        ca-certificates \
+        git \
+        ssh \
+        sudo \
+        wget \
+        xterm \
+        xauth > /dev/null && \
+        rm -rf /var/lib/apt/lists/*
+
+# Add sudo support
+RUN echo "%users ALL = (ALL) NOPASSWD: ALL" >> /etc/sudoers
+
+RUN apt-get update -y > /dev/null && \
+    apt-get install --no-install-recommends -y \
+        python3.8 \
+        python3-pip && \
+        rm -rf /var/lib/apt/lists/*
+
+# Register the version in alternatives
+RUN update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.8 1
+# Set python 3.8 as the default python
+RUN update-alternatives --set python3 /usr/bin/python3.8
+
+# Upgrade Python3 pip
+RUN python3 -m pip --no-cache-dir install --upgrade pip
+
+EXPOSE 25000
+RUN apt-get update && apt-get install -y openssh-server && rm -rf /var/lib/apt/lists/*
+RUN mkdir /var/run/sshd
+
+RUN apt-get update && apt-get install -y liblapacke liblapacke-dev && rm -rf /var/lib/apt/lists/*
+
+RUN apt-get update && apt-get install -y libjpeg8-dev && \
+    rm -rf /var/lib/apt/lists/*
+
+# Set up symlink to point to the correct python version
+RUN ln -sf /usr/bin/python3.8 /usr/bin/python
+RUN ln -s /usr/lib/x86_64-linux-gnu/libjpeg.so /usr/lib
+
+RUN sed -i 's/PermitRootLogin prohibit-password/PermitRootLogin yes/' /etc/ssh/sshd_config && \
+    sed -i 's/Port 22/Port 25000/' /etc/ssh/sshd_config
+
+# SSH login fix. Otherwise user is kicked off after login
+RUN sed 's@session\s*required\s*pam_loginuid.so@session optional pam_loginuid.so@g' -i /etc/pam.d/sshd
+
+# upgrade pip
+RUN python3 -m pip --no-cache-dir install --upgrade pip
+
+# Install the AIMET package wheel files
+COPY *.whl /tmp/
+RUN cd /tmp && python3 -m pip install *.whl -f https://download.pytorch.org/whl/torch_stable.html && rm -f /tmp/*.whl
+
+# Remove onnxruntime install in order to fix onnxruntime-gpu
+RUN export ONNXRUNTIME_VER=$(python3 -c 'import onnxruntime; print(onnxruntime.__version__)') && \
+    python3 -m pip uninstall -y onnxruntime && \
+    python3 -m pip --no-cache-dir install onnxruntime-gpu==$ONNXRUNTIME_VER
+

packaging/dockers/README.md

@@ -0,0 +1,137 @@
+AIMET Docker creation
+=====================
+
+This page provides instructions to build a docker image with AIMET packages and start the development docker container.
+
+Setup workspace
+---------------
+
+```console
+WORKSPACE="<absolute_path_to_workspace>"
+mkdir $WORKSPACE && cd $WORKSPACE
+git clone https://github.com/quic/aimet.git
+cd aimet/packaging/dockers
+```
+
+Make sure no wheel file is present in present working directory
+```console
+rm -rf *.whl
+```
+
+Set variant
+------------
+
+Set the *<variant_string>* to ONE of the following depending on your desired variant
+
+* For the PyTorch 1.13 GPU variant, use **torch_gpu**
+* For the PyTorch 1.13 CPU variant, use **torch_cpu**
+* For the PyTorch 1.9 GPU variant, use **torch_gpu_pt19**
+* For the PyTorch 1.9 CPU variant, use **torch_cpu_pt19**
+* For the TensorFlow GPU variant, use **tf_gpu**
+* For the TensorFlow CPU variant, use **tf_cpu**
+* For the ONNX GPU variant, use **onnx_gpu**
+* For the ONNX CPU variant, use **onnx_cpu**
+
+```console
+export AIMET_VARIANT=<variant_string>
+```
+
+Download AIMET packages
+------------------------
+
+Go to https://github.com/quic/aimet/releases and identify the release tag of the package you want to install.
+
+
+Replace <release_tag> in the steps below with the appropriate tag:
+
+```console
+export release_tag=<release_tag>
+```
+
+Set the package download URL as follows:
+
+```console
+export download_url="https://github.com/quic/aimet/releases/download/${release_tag}"
+```
+
+Set the common suffix for the package files as follows:
+
+```console
+export wheel_file_suffix="cp38-cp38-linux_x86_64.whl"
+```
+
+Download the AIMET packages in the order specified below:
+
+```console
+wget ${download_url}/AimetCommon-${AIMET_VARIANT}_${release_tag}-${wheel_file_suffix}
+
+
+# Download ONE of the following depending on the variant
+wget ${download_url}/AimetTorch-${AIMET_VARIANT}_${release_tag}-${wheel_file_suffix}
+
+# OR
+
+wget ${download_url}/AimetTensorflow-${AIMET_VARIANT}_${release_tag}-${wheel_file_suffix}
+
+# OR
+
+wget ${download_url}/AimetOnnx-${AIMET_VARIANT}_${release_tag}-${wheel_file_suffix}
+
+
+wget ${download_url}/Aimet-${AIMET_VARIANT}_${release_tag}-${wheel_file_suffix}
+```
+
+Build docker image
+------------------
+
+Follow these instructions in order to build the docker image locally. If not, skip to the next section.
+
+```console
+docker_image_name="aimet-prod-docker-${AIMET_VARIANT}:<any_tag>"
+docker_container_name="aimet-prod-${AIMET_VARIANT}-<any_name>"
+
+docker build -t ${docker_image_name} -f Dockerfile.${AIMET_VARIANT} .
+```
+
+**NOTE:** Feel free to modify the *docker_image_name* and *docker_container_name* as needed.
+
+Start docker container
+-----------------------
+
+Ensure that a docker named *$docker_container_name* is not already running; otherwise remove the existing container and then start a new container as follows:
+
+```console
+docker ps -a | grep ${docker_container_name} && docker kill ${docker_container_name}
+
+docker run --rm -it -u $(id -u ${USER}):$(id -g ${USER}) \
+-v /etc/passwd:/etc/passwd:ro -v /etc/group:/etc/group:ro \
+-v ${HOME}:${HOME} -v ${WORKSPACE}:${WORKSPACE} \
+-v "/local/mnt/workspace":"/local/mnt/workspace" \
+--entrypoint /bin/bash -w ${WORKSPACE} --hostname ${docker_container_name} ${docker_image_name}
+```
+
+**NOTE:**
+* Feel free to modify the above *docker run* command based on the environment and filesystem on your host machine.
+* If nvidia-docker 2.0 is installed, then add *--gpus all* to the *docker run* commands in order to enable GPU access inside the docker container.
+* If nvidia-docker 1.0 is installed, then replace *docker run* with *nvidia-docker run* in order to enable GPU access inside the docker container.
+* Port forwarding needs to be done in order to run the Visualization APIs from docker container. This can be achieved by running the docker container as follows:
+
+```console
+
+port_id="<any-port-number>"
+
+docker run -p ${port_id}:${port_id} --rm -it -u $(id -u ${USER}):$(id -g ${USER}) \
+-v /etc/passwd:/etc/passwd:ro -v /etc/group:/etc/group:ro \
+-v ${HOME}:${HOME} -v ${WORKSPACE}:${WORKSPACE} \
+-v "/local/mnt/workspace":"/local/mnt/workspace" \
+--entrypoint /bin/bash -w ${WORKSPACE} --hostname ${docker_container_name} ${docker_image_name}
+```
+
+Environment setup
+------------------
+
+Set the common environment variables as follows:
+
+```console
+source /usr/local/lib/python3.8/dist-packages/aimet_common/bin/envsetup.sh
+```