Lablup FF 21.02 python 3.8 & CUDA 11.1

Author: hephaex

python-ff/Dockerfile.21.02-py38-cuda11.1

@@ -0,0 +1,189 @@
+FROM lablup/common-base:py38-cuda11.1
+
+RUN python3 -m pip install --no-cache-dir --upgrade \
+	    -q git+https://github.com/google-research/tensorflow_constrained_optimization
+	    
+RUN python3 -m pip install --no-cache-dir \
+      --use-feature=2020-resolver \
+        requests_oauthlib==1.2.0 \
+	neural-structured-learning==1.3.1 \
+	pydot==1.4.0 \
+	proto-plus==1.10.2
+RUN python3 -m pip install --no-cache-dir \
+      --use-feature=2020-resolver \
+	    tensorflow-transform==0.25.0 \
+	    tensorflow_model_analysis \
+	    tensorflow-datasets \
+	    tensorflow_probability==0.11.1 \ 
+	    tensorflow-addons==0.11.2 \
+	    neural-structured-learning==1.3.1 \
+	    tensorflow-model-optimization==0.5.0 \
+	    tfx==0.25.0 \
+	    tfx-bsl==0.25.0 \ 
+	    tf-agents==0.6.0  
+
+RUN python3 -m pip install --no-cache-dir \
+      --use-feature=2020-resolver \
+	    fastavro==1.2.0 \
+	    pymongo==3.11.2 \
+	    docker==4.4.1 \
+	    pbr==5.5.1 \
+	    mock==1.3.0 \
+	    proto-plus==1.13.0 \
+	    oauth2client==4.1.3 \
+	    future==0.18.1 \
+	    gin_config==0.4.0 \
+	    joblib==0.14.0 \
+	    libcst==0.3.14 
+
+RUN python3 -m pip install --no-cache-dir \
+      --use-feature=2020-resolver \
+	    google_cloud_spanner==1.19.0 \
+	    googleapis-common-protos==1.51.0 \
+	    google_cloud_videointelligence==1.15.0 \
+	    google_cloud_vision==0.42.0 \
+	    google_cloud_pubsub==1.6.0 \
+	    google_cloud_dlp==0.15.0 \
+	    google_cloud_datastore==1.15.2 \
+	    google_cloud_bigtable==1.6.0 \
+	    google_api_core==1.24.0 
+
+RUN python3 -m pip install --no-cache-dir \
+	    mesh-tensorflow==0.1.18 \
+            tensorflow-cloud==0.1.10 \
+            tensorflow-hub==0.9.0 
+
+RUN python3 -m pip install --no-cache-dir \
+	    -i https://pypi-nightly.tensorflow.org/simple tensorflow-data-validation
+RUN python3 -m pip install --no-cache-dir \
+	    tensorflow-graphics-gpu==1.0.0 \
+            tensorflow-io==0.17.0 \
+	    tensorflow_ranking==0.3.2 \
+	    tensorflow-plot==0.3.2 \
+	    tensorflow_text==2.4.1 \
+	    tensorflow-gan==2.0.0 
+RUN python3 -m pip uninstall --no-cache-dir -y \
+            tensorflow==2.3.0 \
+	    tensorflow-gpu==2.4.1 \
+	    tensorboard==2.3.0 \
+	    tensorboard-plugin-wit==1.7.0 && \
+    python3 -m pip install --no-cache-dir \
+            tensorflow-gpu==2.4.1 \
+	    tensorboard-plugin-wit==1.8.0 \
+	    tensorboard==2.4.1
+
+RUN python3 -m pip install --extra-index-url \
+       https://developer.download.nvidia.com/compute/redist nvidia-dali-cuda100 \
+       && \
+    python3 -m pip install --extra-index-url \
+       https://developer.download.nvidia.com/compute/redist nvidia-dali-tf-plugin-cuda100
+
+# Install PyTorch and MXNet
+ENV PYTORCH_VERSION=1.7.1
+ENV TORCHVISION_VERSION=0.8.2
+ENV TORCHAUDIO_VERSION=0.7.2
+ENV TORCHTEXT_VERSION=0.8.1
+ENV TENSORBOARDX_VERSION=2.1
+ENV MXNET_VERSION=1.6.0
+
+RUN python3 -m pip uninstall -y torch && \
+    python3 -m pip install --no-cache-dir \
+    	https://download.pytorch.org/whl/cu110/torch-${PYTORCH_VERSION}%2Bcu110-cp38-cp38-linux_x86_64.whl \
+	https://download.pytorch.org/whl/cu110/torchvision-${TORCHVISION_VERSION}%2Bcu110-cp38-cp38-linux_x86_64.whl \
+        https://download.pytorch.org/whl/torchaudio-${TORCHAUDIO_VERSION}-cp38-cp38-linux_x86_64.whl \
+        https://download.pytorch.org/whl/torchtext-${TORCHTEXT_VERSION}-cp38-cp38-linux_x86_64.whl && \
+    python3 -m pip install --no-cache-dir tensorboardX==${TENSORBOARDX_VERSION}
+
+RUN python3 -m pip install --no-cache-dir mxnet mkl
+
+# Install Horovod, temporarily using CUDA stubs
+RUN ldconfig /usr/local/cuda/targets/x86_64-linux/lib/stubs && \
+    HOROVOD_GPU_ALLREDUCE=NCCL HOROVOD_GPU_BROADCAST=NCCL \
+    HOROVOD_WITH_TENSORFLOW=1 HOROVOD_WITH_PYTORCH=1 HOROVOD_WITHOUT_MXNET=1 \
+    pip install --no-cache-dir horovod==0.21.1 && \
+    ldconfig
+
+RUN python3 -m pip install --no-cache-dir \
+    	    mpi4py==3.0.3 \
+            mlflow==1.12.1 \
+	    nni==1.9 \
+	    scikit-nni==0.2.1
+
+# install kaldi
+WORKDIR /opt
+RUN apt update && \
+    apt list --upgradable && \
+    apt-get install -y --no-install-recommends \
+	patch wget rsync jq gnupg libtcmalloc-minimal4 && \
+    apt-get install -y --no-install-recommends \
+        automake \
+        autoconf \
+        flac \
+        gawk \
+        libatlas3-base \
+        libtool \
+        sox \
+	subversion \
+        unzip \
+        bc \
+        libatlas-base-dev \
+        gfortran \
+        zlib1g-dev && \
+    apt-get install -y --no-install-recommends \
+        git libglib2.0-0 less libnl-route-3-200 libnl-3-dev \
+	libnl-route-3-dev libnuma-dev libnuma1 \
+	libpmi2-0-dev nano numactl openssh-client vim && \
+    rm -rf /var/lib/apt/lists/*
+RUN ln -sf /usr/include/x86_64-linux-gnu/atlas /usr/local/include/atlas && \
+    ln -sf /usr/include/x86_64-linux-gnu/cblas.h /usr/local/include/cblas.h && \
+    ln -sf /usr/include/x86_64-linux-gnu/clapack.h /usr/local/include/clapack.h && \
+    ln -sf /usr/lib/x86_64-linux-gnu/atlas /usr/local/lib/atlas
+
+RUN git clone https://github.com/hephaex/kaldi.git /opt/kaldi && \
+    cd /opt/kaldi/tools && \
+    make -j $(nproc) && \
+    cd /opt/kaldi/src && \
+    ./configure --shared --use-cuda \
+	--cudatk-dir=/usr/local/cuda/ \
+	--mathlib=ATLAS \
+    	--cuda-arch="-gencode arch=compute_52,code=sm_52 -gencode arch=compute_60,code=sm_60 -gencode arch=compute_61,code=sm_61 -gencode arch=compute_70,code=sm_70 -gencode arch=compute_75,code=sm_75 -gencode arch=compute_80,code=sm_80" && \
+    make depend -j $(nproc) && \
+    make -j $(nproc) 
+RUN cd /opt/kaldi/src  && \
+    make ext -j $(nproc) && \
+    ldconfig && \
+    find /opt/kaldi  -type f \( -name "*.o" -o -name "*.la" -o -name "*.a" \) -\
+exec rm {} \; && \
+    rm -rf /opt/kaldi/.git
+
+RUN mkdir -p /workspace && \
+    chmod -R 777 /workspace
+
+RUN apt autoclean && \
+    rm -rf /var/lib/apt/lists/* && \	
+    rm -rf /root/.cache && \
+    rm -rf /tmp/*
+
+# Copy Backend.Ai multi-node support
+COPY runner-scripts/bootstrap.sh runner-scripts/setup_multinode.py /opt/container/
+COPY service-defs /etc/backend.ai/service-defs
+
+# Install ipython kernelspec
+Run python3 -m ipykernel install --display-name "Lablup FF 21.02 on Python 3.8 (CUDA 11.1)" && \
+    cat /usr/local/share/jupyter/kernels/python3/kernel.json
+
+# Backend.AI specifics
+LABEL ai.backend.kernelspec="1" \
+      ai.backend.envs.corecount="OPENBLAS_NUM_THREADS,OMP_NUM_THREADS,NPROC" \
+      ai.backend.features="batch query uid-match user-input" \
+      ai.backend.base-distro="ubuntu16.04" \
+      ai.backend.resource.min.cpu="1" \
+      ai.backend.resource.min.mem="1g" \
+      ai.backend.resource.min.cuda.device=0 \
+      ai.backend.resource.min.cuda.shares=0 \
+      ai.backend.runtime-type="python" \
+      ai.backend.runtime-path="/usr/bin/python3" \
+      ai.backend.service-ports="ipython:pty:3000,jupyter:http:8081,jupyterlab:http:8090,vscode:http:8180,tensorboard:http:6006,mlflow-ui:preopen:5000,nniboard:preopen:8080"
+
+WORKDIR /home/work
+# vim: ft=dockerfile