Skip to content

Commit 70ab782

Browse files
hephaexhephaex
committed
update NCCL 2.7.6
1 parent 13379b8 commit 70ab782

File tree

2 files changed

+122
-1
lines changed

2 files changed

+122
-1
lines changed

‎commons/Dockerfile.base.20.08-py36-cuda10.1

Lines changed: 2 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -5,7 +5,7 @@ ARG CUDA=10.1
55
ARG LIBNVINFER=6.0.1-1
66
ARG LIBNVINFER_MAJOR_VERSION=6
77
ARG CUDNN=7.6.5.32-1
8-
ENV NCCL_VERSION=2.7.3-1+cuda10.1
8+
ENV NCCL_VERSION=2.7.6-1+cuda10.1
99

1010
# Python 2.7 or 3.6 is supported by Ubuntu Bionic out of the box
1111
ARG python=3.6
@@ -358,6 +358,7 @@ RUN jupyter nbextensions_configurator enable && \
358358
jupyter serverextension enable --py jupyter_lsp && \
359359
jupyter labextension install --no-build @jupyterlab/toc && \
360360
# jupyter labextension install @jupyterlab/hdf5
361+
jupyter nbextension enable --py tensorflow_model_analysis \
361362
jupyter labextension install --no-build jupyterlab-nvdashboard && \
362363
jupyter lab build
363364

‎python-ff/Dockerfile.20.08-py36-cuda10.1

Lines changed: 120 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,120 @@
1+
FROM lablup/common-base:20.08-py36-cuda10.1
2+
3+
# Release date: 2020.07.26
4+
# Install TensorFlow, Keras, PyTorch and MXNet
5+
ENV TENSORFLOW_VERSION=2.2.0
6+
ENV PYTORCH_VERSION=1.5.1
7+
ENV TORCHVISION_VERSION=0.6.1
8+
ENV TORCHAUDIO_VERSION=0.5.1
9+
ENV TORCHTEXT_VERSION=0.6.0
10+
ENV TENSORBOARDX_VERSION=2.1
11+
ENV MXNET_VERSION=1.6.0
12+
ARG CUDA=10.1
13+
ARG LIBNVINFER=6.0.1-1
14+
ARG LIBNVINFER_MAJOR_VERSION=6
15+
ENV CUDNN_VERSION=7.6.5.32-1+cuda10.1
16+
ENV NCCL_VERSION=2.7.6-1+cuda10.1
17+
ENV DEBIAN_FRONTEND=noninteractive
18+
ENV mecab_dicdir /usr/local/lib/mecab/dic/mecab-ko-dic
19+
ENV MPLBACKEND=Svg
20+
ENV PYTHONUNBUFFERED=1 \
21+
LD_LIBRARY_PATH="/usr/local/cuda/extras/CUPTI/lib64:/usr/local/cuda/lib64:/usr/local/nvidia/lib64:/usr/include/x86_64-linux-gnu" \
22+
PATH="/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/local/bin/mecab" \
23+
LANG=C.UTF-8
24+
25+
# Set default shell to /bin/bash
26+
SHELL ["/bin/bash", "-cu"]
27+
28+
RUN python3 -m pip install --no-cache-dir --upgrade \
29+
gast==0.3.3 \
30+
'httplib2<0.18.0,>=0.8' \
31+
'coverage<5.0,>=4.5' \
32+
mock==3.0.5 \
33+
tensorflow==${TENSORFLOW_VERSION} \
34+
mesh-tensorflow==0.1.16 \
35+
tensorboard==2.2.2
36+
37+
RUN python3 -m pip install --no-cache-dir --upgrade \
38+
-q git+https://github.com/google-research/tensorflow_constrained_optimization
39+
RUN python3 -m pip install --no-cache-dir --upgrade \
40+
cachetools==3.1.0 \
41+
tensorflow-transform==0.22.0 \
42+
tensorflow_model_analysis==0.22.2 \
43+
ngraph-tensorflow-bridge==0.18.0 \
44+
tensorflow-metadata==0.22.2 \
45+
tensorflow-datasets==3.2.1 \
46+
tensorflow_probability==0.10.1 \
47+
tensorwatch==0.9.1 \
48+
tensorflow-hub==0.8.0 \
49+
tensorflow-addons==0.10.0 \
50+
tensorflow_text==2.2.1 \
51+
neural-structured-learning==1.2.0 \
52+
tensorflow_constrained_optimization==0.2 \
53+
tensorflow-graphics-gpu==1.0.0 \
54+
tensorflow-gan==2.0.0 \
55+
tensorflow-data-validation==0.22.2 \
56+
tensorflow-model-optimization==0.3.0
57+
RUN python3 -m pip install --no-cache-dir --upgrade \
58+
jupyter-tensorboard==0.2.0 \
59+
tf-agents==0.5.0 \
60+
tf-slim==1.1.0 \
61+
tensorflow-plot==0.3.2 \
62+
tensorflow-lattice==2.0.5 \
63+
tensorflow-io==0.14.0 \
64+
tfx==0.22.1 \
65+
tfx-bsl==0.22.1
66+
RUN python3 -m pip install --no-cache-dir \
67+
tensorflow_ranking==0.3.1 \
68+
tensorflow-compression==1.3 \
69+
tensor2tensor==1.15.7 \
70+
jupyterlab-nvdashboard==0.3.1
71+
72+
RUN python3 -m pip install --no-cache-dir --upgrade \
73+
-q git+https://github.com/tensorflow/fairness-indicators
74+
75+
RUN python3 -m pip install pip --no-cache-dir \
76+
https://download.pytorch.org/whl/cu101/torch-${PYTORCH_VERSION}%2Bcu101-cp36-cp36m-linux_x86_64.whl \
77+
https://download.pytorch.org/whl/cu101/torchvision-${TORCHVISION_VERSION}%2Bcu101-cp36-cp36m-linux_x86_64.whl \
78+
https://download.pytorch.org/whl/torchaudio-${TORCHAUDIO_VERSION}-cp36-cp36m-linux_x86_64.whl \
79+
torchtext==${TORCHTEXT_VERSION} && \
80+
python3 -m pip install --no-cache-dir tensorboardX==${TENSORBOARDX_VERSION}
81+
82+
RUN python3 -m pip install --no-cache-dir mxnet-cu101mkl==${MXNET_VERSION}
83+
84+
RUN python3 -m pip install --no-cache-dir --extra-index-url \
85+
https://developer.download.nvidia.com/compute/redist \
86+
nvidia-dali-cuda100 && \
87+
python3 -m pip install --no-cache-dir --extra-index-url \
88+
https://developer.download.nvidia.com/compute/redist nvidia-dali-tf-plugin-cuda100
89+
90+
# Install Horovod, temporarily using CUDA stubs
91+
RUN ldconfig /usr/local/cuda/targets/x86_64-linux/lib/stubs && \
92+
HOROVOD_GPU_ALLREDUCE=NCCL HOROVOD_GPU_BROADCAST=NCCL \
93+
HOROVOD_WITH_TENSORFLOW=1 HOROVOD_WITH_PYTORCH=1 HOROVOD_WITH_MXNET=1 \
94+
pip install --no-cache-dir horovod==0.19.5 && \
95+
ldconfig
96+
97+
RUN python3 -m pip install --no-cache-dir --upgrade \
98+
mpi4py==3.0.3 \
99+
nni==1.7 \
100+
scikit-nni==0.2.1
101+
102+
# Install ipython kernelspec
103+
Run python3 -m ipykernel install --display-name "Lablup FF 20.08 on Python 3.6 (CUDA 10.1)" && \
104+
cat /usr/local/share/jupyter/kernels/python3/kernel.json
105+
106+
# Backend.AI specifics
107+
LABEL ai.backend.kernelspec="1" \
108+
ai.backend.envs.corecount="OPENBLAS_NUM_THREADS,OMP_NUM_THREADS,NPROC" \
109+
ai.backend.features="batch query uid-match user-input" \
110+
ai.backend.base-distro="ubuntu16.04" \
111+
ai.backend.resource.min.cpu="1" \
112+
ai.backend.resource.min.mem="1g" \
113+
ai.backend.resource.min.cuda.device=0 \
114+
ai.backend.resource.min.cuda.shares=0 \
115+
ai.backend.runtime-type="python" \
116+
ai.backend.runtime-path="/usr/bin/python3" \
117+
ai.backend.service-ports="ipython:pty:3000,jupyter:http:8080,jupyterlab:http:8090,vscode:http:8180,tensorboard:http:6006"
118+
119+
WORKDIR /home/work
120+
# vim: ft=dockerfile

0 commit comments

Comments
 (0)