华为人工智能atlas A800-9000 物理服务器全程离线安装驱动以及CANN安装部署和MindSpore安装部署和Tensorflow安装部署
A800-9000 物理服务器安装驱动
使用镜像配置本地apt源
root@ubuntu:/etc/apt# mkdir /media/cdrom
root@ubuntu:/etc/apt# mount /home/cby/ubuntu-18.04.5-server-arm64.iso /media/cdrom
mount: /media/chrom: WARNING: device write-protected, mounted read-only.
root@ubuntu:/etc/apt# apt-cdrom -m -d=/media/cdrom/ add
root@ubuntu:/etc/apt# cat /etc/apt/sources.list
创建普通用户并设置密码
root@ubuntu:~# groupadd HwHiAiUser
root@ubuntu:~# useradd -g HwHiAiUser -d /home/HwHiAiUser -m HwHiAiUser
root@ubuntu:~# passwd HwHiAiUser
Enter new UNIX password:
Retype new UNIX password:
passwd: password updated successfully
安装驱动以及固件
root@ubuntu:~# cd /home/cby/
root@ubuntu:/home/cby# ll
total 98324
drwxr-xr-x 4 cby cby 4096 Apr 21 21:41 ./
drwxr-xr-x 4 root root 4096 Apr 21 21:44 ../
-rw-r--r-- 1 cby cby 99728721 Apr 21 21:41 A800-9000-npu-driver_20.2.0_ubuntu18.04-aarch64.run
-rw-r--r-- 1 cby cby 912335 Apr 21 21:41 A800-9000-npu-firmware_1.76.22.3.220.run
root@ubuntu:/home/cby# chmod +x *.run
root@ubuntu:/home/cby# ll
total 98324
drwxr-xr-x 4 cby cby 4096 Apr 21 21:41 ./
drwxr-xr-x 4 root root 4096 Apr 21 21:44 ../
-rwxr-xr-x 1 cby cby 99728721 Apr 21 21:41 A800-9000-npu-driver_20.2.0_ubuntu18.04-aarch64.run*
-rwxr-xr-x 1 cby cby 912335 Apr 21 21:41 A800-9000-npu-firmware_1.76.22.3.220.run*
root@ubuntu:/home/cby# apt install gcc
root@ubuntu:/home/cby# apt install make
root@ubuntu:/home/cby# ./A800-9000-npu-driver_20.2.0_ubuntu18.04-aarch64.run –run
root@ubuntu:/home/cby# ./A800-9000-npu-firmware_1.76.22.3.220.run –run
*注意：安装完成后需要重启服务器
验证是否安装成功
root@ubuntu:/home/cby# npu-smi info
CANN开发环境部署安装
安装环境以及依赖
root@ubuntu:/home/cby# apt install g++
root@ubuntu:/home/cby# cd cmake/
root@ubuntu:/home/cby/cmake# ll
total 4356
drwxr-xr-x 2 root root 4096 Apr 21 23:48 ./
drwxr-xr-x 7 cby cby 4096 Apr 21 23:48 ../
-rw-r--r-- 1 cby cby 2971248 Apr 21 23:45 cmake_3.10.2-1ubuntu2.18.04.1_arm64.deb
-rw-r--r-- 1 cby cby 1331524 Apr 21 23:45 cmake-data_3.10.2-1ubuntu2.18.04.1_all.deb
-rw-r--r-- 1 cby cby 69166 Apr 21 23:47 libjsoncpp1_1.7.4-3_arm64.deb
-rw-r--r-- 1 cby cby 71788 Apr 21 23:48 librhash0_1.3.6-2_arm64.deb
root@ubuntu:/home/cby/cmake# apt install ./*
root@ubuntu:/home/cby/cmake# make –version
root@ubuntu:/home/cby# apt install ./zlib1g-dev_1%3a1.2.11.dfsg-0ubuntu2_arm64.deb
root@ubuntu:/home/cby# apt install ./libbz2-dev_1.0.6-8.1ubuntu0.2_arm64.deb
root@ubuntu:/home/cby# apt install ./libsqlite3-dev_3.22.0-1ubuntu0.4_arm64.deb
root@ubuntu:/home/cby# cd libssl-dev/
root@ubuntu:/home/cby/libssl-dev# apt install ./*
root@ubuntu:/home/cby# cd libxslt1-dev/
root@ubuntu:/home/cby/libxslt1-dev# ll
total 13596
drwxr-xr-x 2 root root 4096 Apr 22 00:37 ./
drwxr-xr-x 10 cby cby 4096 Apr 22 00:37 ../
-rw-r--r-- 1 cby cby 18528 Apr 22 00:30 gir1.2-harfbuzz-0.0_1.7.2-1ubuntu1_arm64.deb
-rw-r--r-- 1 cby cby 170204 Apr 22 00:27 icu-devtools_60.2-3ubuntu3.1_arm64.deb
-rw-r--r-- 1 cby cby 983364 Apr 22 00:37 libglib2.0-0_2.56.4-0ubuntu0.18.04.8_arm64.deb
-rw-r--r-- 1 cby cby 61832 Apr 22 00:33 libglib2.0-bin_2.56.4-0ubuntu0.18.04.8_arm64.deb
-rw-r--r-- 1 cby cby 1297600 Apr 22 00:31 libglib2.0-dev_2.56.4-0ubuntu0.18.04.8_arm64.deb
-rw-r--r-- 1 cby cby 99676 Apr 22 00:31 libglib2.0-dev-bin_2.56.4-0ubuntu0.18.04.8_arm64.deb
-rw-r--r-- 1 cby cby 14528 Apr 22 00:32 libgraphite2-dev_1.3.11-2_arm64.deb
-rw-r--r-- 1 cby cby 280584 Apr 22 00:28 libharfbuzz-dev_1.7.2-1ubuntu1_arm64.deb
-rw-r--r-- 1 cby cby 12556 Apr 22 00:30 libharfbuzz-gobject0_1.7.2-1ubuntu1_arm64.deb
-rw-r--r-- 1 cby cby 5348 Apr 22 00:29 libharfbuzz-icu0_1.7.2-1ubuntu1_arm64.deb
-rw-r--r-- 1 cby cby 8890124 Apr 22 00:26 libicu-dev_60.2-3ubuntu3.1_arm64.deb
-rw-r--r-- 1 cby cby 14412 Apr 22 00:28 libicu-le-hb0_1.0.3+git161113-4_arm64.deb
-rw-r--r-- 1 cby cby 29760 Apr 22 00:27 libicu-le-hb-dev_1.0.3+git161113-4_arm64.deb
-rw-r--r-- 1 cby cby 18756 Apr 22 00:26 libiculx60_60.2-3ubuntu3.1_arm64.deb
-rw-r--r-- 1 cby cby 120696 Apr 22 00:35 libpcre16-3_2%3a8.39-9_arm64.deb
-rw-r--r-- 1 cby cby 113240 Apr 22 00:35 libpcre32-3_2%3a8.39-9_arm64.deb
-rw-r--r-- 1 cby cby 459316 Apr 22 00:33 libpcre3-dev_2%3a8.39-9_arm64.deb
-rw-r--r-- 1 cby cby 15124 Apr 22 00:35 libpcrecpp0v5_2%3a8.39-9_arm64.deb
-rw-r--r-- 1 cby cby 673384 Apr 22 00:25 libxml2-dev_2.9.4+dfsg1-6.1ubuntu1.3_arm64.deb
-rw-r--r-- 1 cby cby 395564 Apr 22 00:24 libxslt1-dev_1.1.29-5ubuntu0.2_arm64.deb
-rw-r--r-- 1 cby cby 42802 Apr 22 00:33 pkg-config_0.29.1-0ubuntu2_arm64.deb
-rw-r--r-- 1 cby cby 144176 Apr 22 00:37 python3-distutils_3.6.9-1~18.04_all.deb
root@ubuntu:/home/cby/libxslt1-dev# apt install ./*
root@ubuntu:/home/cby# cd libffi-dev/
root@ubuntu:/home/cby/libffi-dev# ls
libffi-dev_3.2.1-8_arm64.deb
root@ubuntu:/home/cby/libffi-dev# apt install ./*
root@ubuntu:/home/cby# apt install unzip
root@ubuntu:/home/cby# apt install ./libblas-dev_3.7.1-4ubuntu1_arm64.deb
root@ubuntu:/home/cby# cd gfortran/
root@ubuntu:/home/cby/gfortran# ll
total 7844
drwxr-xr-x 2 root root 4096 Apr 22 00:50 ./
drwxr-xr-x 12 cby cby 4096 Apr 22 00:50 ../
-rw-r--r-- 1 cby cby 1344 Apr 22 00:48 gfortran_4%3a7.4.0-1ubuntu2.3_arm64.deb
-rw-r--r-- 1 cby cby 7464740 Apr 22 00:48 gfortran-7_7.5.0-3ubuntu1~18.04_arm64.deb
-rw-r--r-- 1 cby cby 248176 Apr 22 00:50 libgfortran4_7.5.0-3ubuntu1~18.04_arm64.deb
-rw-r--r-- 1 cby cby 300500 Apr 22 00:49 libgfortran-7-dev_7.5.0-3ubuntu1~18.04_arm64.deb
root@ubuntu:/home/cby/gfortran# apt install ./*
root@ubuntu:/home/cby# cd libblas3/
root@ubuntu:/home/cby/libblas3# apt install ./libblas3_3.7.1-4ubuntu1_arm64.deb
root@ubuntu:/home/cby# cd libopenblas-dev/
root@ubuntu:/home/cby/libopenblas-dev# ll
total 3412
drwxr-xr-x 2 root root 4096 Apr 22 00:56 ./
drwxr-xr-x 14 cby cby 4096 Apr 22 00:56 ../
-rw-r--r-- 1 cby cby 1813748 Apr 22 00:55 libopenblas-base_0.2.20+ds-4_arm64.deb
-rw-r--r-- 1 cby cby 1668126 Apr 22 00:54 libopenblas-dev_0.2.20+ds-4_arm64.deb
root@ubuntu:/home/cby/libopenblas-dev# apt install ./*
安装完成后查看版本
gcc --version
g++ --version
make --version
cmake --version
dpkg -l zlib1g| grep zlib1g| grep ii
dpkg -l zlib1g-dev| grep zlib1g-dev| grep ii
dpkg -l libbz2-dev| grep libbz2-dev| grep ii
dpkg -l libsqlite3-dev| grep libsqlite3-dev| grep ii
dpkg -l openssl| grep openssl| grep ii
dpkg -l libssl-dev| grep libssl-dev| grep ii
dpkg -l libxslt1-dev| grep libxslt1-dev| grep ii
dpkg -l libffi-dev| grep libffi-dev| grep ii
dpkg -l unzip| grep unzip| grep ii
dpkg -l pciutils| grep pciutils| grep ii
dpkg -l net-tools| grep net-tools| grep ii
dpkg -l libblas-dev| grep libblas-dev| grep ii
dpkg -l gfortran| grep gfortran| grep ii
dpkg -l libblas3| grep libblas3| grep ii
dpkg -l libopenblas-dev| grep libopenblas-dev| grep ii
安装Python3.7.5
root@ubuntu:/home/cby/python# tar xvf Python3.7.5.tar
root@ubuntu:/home/cby/python# cd Python-3.7.5/
root@ubuntu:/home/cby/python/Python-3.7.5# ./configure --prefix=/usr/local/python3.7.5 --enable-loadable-sqlite-extensions --enable-shared
root@ubuntu:/home/cby/python/Python-3.7.5# make
root@ubuntu:/home/cby/python/Python-3.7.5# make install
root@ubuntu:/home/cby# sudo ln -s /usr/local/python3.7.5/bin/pip3 /usr/local/bin/pip3.7.5
root@ubuntu:/home/cby# sudo ln -s /usr/local/python3.7.5/bin/python3 /usr/local/bin/python3.7.5
root@ubuntu:/home/cby/cann_xunlian# sudo ln -s /usr/local/python3.7.5/bin/python3 /usr/local/bin/python3.7
root@ubuntu:/home/cby/cann_xunlian# sudo ln -s /usr/local/python3.7.5/bin/pip3 /usr/local/bin/pip3.7
root@ubuntu:/home/cby# vim ~/.bashrc
export LD_LIBRARY_PATH=/usr/local/python3.7.5/lib:$LD_LIBRARY_PATH
root@ubuntu:/home/cby# python3.7.5 --version
Python 3.7.5
root@ubuntu:/home/cby# pip3.7.5 --version
pip 19.2.3 from /usr/local/python3.7.5/lib/python3.7/site-packages/pip (python 3.7)
使用Python3.7.5环境安装pip依赖包
root@ubuntu:/home/cby/pip-pack# tar xvf pip_pack.tar
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./attrs-20.3.0-py2.py3-none-any.whl
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./numpy-1.17.2-cp37-cp37m-linux_aarch64.whl
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./decorator-5.0.6-py3-none-any.whl
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./mpmath-1.2.1-py3-none-any.whl
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./sympy-1.4-py2.py3-none-any.whl
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./pycparser-2.20-py2.py3-none-any.whl
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./cffi-1.12.3.tar.gz
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./PyYAML-5.3.1.tar.gz
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./six-1.15.0-py2.py3-none-any.whl
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./pathlib2-2.3.5-py2.py3-none-any.whl
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./psutil-5.8.0.tar.gz
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./protobuf-3.15.8-py2.py3-none-any.whl
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./scipy-1.6.0-cp37-cp37m-linux_aarch64.whl
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./chardet-3.0.4-py2.py3-none-any.whl
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./idna-2.10-py2.py3-none-any.whl
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./urllib3-1.25.10-py2.py3-none-any.whl
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./certifi-2020.6.20-py2.py3-none-any.whl
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./certifi-2020.6.20-py2.py3-none-any.whl
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./requests-2.24.0-py2.py3-none-any.wh
root@ubuntu:/home/cby/pip-pack/pip_pack# pip3.7.5 install ./xlrd-1.2.0-py2.py3-none-any.whl
*注意：以上pip包的安装必须以该顺序依次进行安装
安装开发套件包
root@ubuntu:/home/cby/cann# ./Ascend-cann-tfplugin_20.2.rc1_linux-aarch64.run –install
root@ubuntu:/home/cby/cann# ./Ascend-cann-toolkit_20.2.rc1_linux-aarch64.run –install
出现install success后表示安装成功。
CANN训练环境部署安装
说明
训练环境的Python3.7.5和环境以及依赖，和开发环境下的安装方式一样，可参考《CANN开发环境部署安装》文档进行安装。在已经搭建好的开发环境中，进行安装训练环境仅需安装一下训练软件包和实用工具包即可。
安装训练软件包
root@ubuntu:/home/cby/cann_xunlian# chmod +x ./*.run
root@ubuntu:/home/cby/cann_xunlian# ./Ascend-cann-nnae_20.2.rc1_linux-aarch64.run –install
root@ubuntu:/home/cby/cann_xunlian# ./Ascend-cann-toolbox_20.2.rc1_linux-aarch64.run –install
出现install success后表示安装成功。
安装MindSpore
安装whl包
安装Ascend 910 AI处理器配套软件包提供的whl包，whl包随配套软件包发布，升级配套软件包之后需要重新安装。
root@ubuntu:/home/cby/mindspore_ascend# pip3.7.5 install /usr/local/Ascend/ascend-toolkit/latest/fwkacllib/lib64/hccl-0.1.0-py3-none-any.whl
root@ubuntu:/home/cby/mindspore_ascend# pip3.7.5 install /usr/local/Ascend/ascend-toolkit/latest/fwkacllib/lib64/te-0.4.0-py3-none-any.whl
root@ubuntu:/home/cby/mindspore_ascend# pip3.7.5 install /usr/local/Ascend/ascend-toolkit/latest/fwkacllib/lib64/topi-0.4.0-py3-none-any.whl
root@ubuntu:/home/cby/mindspore_ascend/pip# pip3.7.5 install easydict-1.9.tar.gz
root@ubuntu:/home/cby/mindspore_ascend/pip# pip3.7.5 install ./wheel-0.36.2-py2.py3-none-any.whl
root@ubuntu:/home/cby/mindspore_ascend/pip# pip3.7.5 install ./astunparse-1.6.3-py2.py3-none-any.whl
root@ubuntu:/home/cby/mindspore_ascend/pip# pip3.7.5 install ./Pillow-8.2.0-cp37-cp37m-linux_aarch64.whl
root@ubuntu:/home/cby/mindspore_ascend/pip# pip3.7.5 install ./asttokens-2.0.4-py2.py3-none-any.whl
root@ubuntu:/home/cby/mindspore_ascend/pip# pip3.7.5 install ./cffi-1.14.5-cp37-cp37m-linux_aarch64.whl
root@ubuntu:/home/cby/mindspore_ascend/pip# pip3.7.5 install ./pyparsing-2.4.7-py2.py3-none-any.whl
root@ubuntu:/home/cby/mindspore_ascend/pip# pip3.7.5 install ./packaging-20.9-py2.py3-none-any.whl
root@ubuntu:/home/cby/mindspore_ascend/pip# pip3.7.5 install ../mindspore_ascend-1.1.1-cp37-cp37m-linux_aarch64.whl
*注意：安装时必须以此顺序进行安装
配置环境变量
# control log level. 0-DEBUG, 1-INFO, 2-WARNING, 3-ERROR, default level is WARNING.
export GLOG_v=2
# Conda environmental options
LOCAL_ASCEND=/usr/local/Ascend
# the root directory of run package
# lib libraries that the run package depends on
export LD_LIBRARY_PATH=${LOCAL_ASCEND}/add-ons/:${LOCAL_ASCEND}/ascend-toolkit/latest/fwkacllib/lib64:${LOCAL_ASCEND}/driver/lib64:${LOCAL_ASCEND}/opp/op_impl/built-in/ai_core/tbe/op_tiling:${LD_LIBRARY_PATH}
# Environment variables that must be configured
export TBE_IMPL_PATH=${LOCAL_ASCEND}/ascend-toolkit/latest/opp/op_impl/built-in/ai_core/tbe
# TBE operator implementation tool path
export ASCEND_OPP_PATH=${LOCAL_ASCEND}/ascend-toolkit/latest/opp
# OPP path
export PATH=${LOCAL_ASCEND}/ascend-toolkit/latest/fwkacllib/ccec_compiler/bin/:${PATH}
# TBE operator compilation tool path
export PYTHONPATH=${TBE_IMPL_PATH}:${PYTHONPATH}
# Python library that TBE implementation depends on
测试是否可行
Python代码内容：
import numpy as np
from mindspore import Tensor
import mindspore.ops as ops
import mindspore.context as context
context.set_context(device_target="Ascend")
x = Tensor(np.ones([1,3,3,4]).astype(np.float32))
y = Tensor(np.ones([1,3,3,4]).astype(np.float32))
print(ops.tensor_add(x, y))
出现此结果即是安装部署完成
[[[[2. 2. 2. 2.]
[2. 2. 2. 2.]
[2. 2. 2. 2.]]
[[2. 2. 2. 2.]
[2. 2. 2. 2.]
[2. 2. 2. 2.]]
[[2. 2. 2. 2.]
[2. 2. 2. 2.]
[2. 2. 2. 2.]]]]
安装mindinsight
安装whl包
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./itsdangerous-1.1.0-py2.py3-none-any.whl
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./Werkzeug-1.0.1-py2.py3-none-any.whl
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./MarkupSafe-1.1.1-cp37-cp37m-linux_aarch64.whl
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./Jinja2-2.11.3-py2.py3-none-any.whl
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./click-7.1.2-py2.py3-none-any.whl
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./Flask-1.1.2-py2.py3-none-any.whl
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./Flask_Cors-3.0.10-py2.py3-none-any.whl
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./yapf-0.31.0-py2.py3-none-any.whl
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./future-0.18.2.tar.gz
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./treelib-1.6.1.tar.gz
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./grpcio-1.37.0-cp37-cp37m-linux_aarch64.whl
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./google_pasta-0.2.0-py3-none-any.whl
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./pytz-2021.1-py2.py3-none-any.whl
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./python_dateutil-2.8.1-py2.py3-none-any.whl
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./pandas-1.2.3-cp37-cp37m-linux_aarch64.whl
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./gunicorn-20.1.0.tar.gz
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./marshmallow-3.11.1-py2.py3-none-any.whl
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./threadpoolctl-2.1.0-py3-none-any.whl
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./joblib-1.0.1-py3-none-any.whl
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./scikit_learn-0.24.1-cp37-cp37m-linux_aarch64.whl
root@ubuntu:/home/cby/mindinsight/Mindinsight# pip3.7.5 install ./mindinsight-1.1.1-cp37-cp37m-linux_aarch64.whl
*注意：安装必须以此顺序进安装
配置环境变量
在配置文件中配置如下变量
PATH=$PATH:/usr/local/python3.7.5/bin/
root@ubuntu:/home/cby# source /etc/profile
启动及使用
root@ubuntu:/home/cby# mindinsight start
Workspace: /root/mindinsight
Web address: http://127.0.0.1:8080
service start state: success
出现该消息后，说明可视化已经启动成功，若需要外机访问的话，需要进行反向代理到0.0.0.0上面即可，比如frp工具即可实现该操作
在训练完成的Python代码目录下，使以下命令即可启动并展示该目录下的训练数据，debugger的参数可使用false或者true
mindinsight start --summary-base-dir . --port 8080 --enable-debugger True --debugger-port 50051
使用如下命令即可启动训练
root@ubuntu:/home/cby/lenet/lenet# python3.7.5 lenet.py --device_target=Ascend
安装Tensorflow
编译hdf5
root@ubuntu:/home/cby/Tensorflow/Tensorflow# cd hdf5-1.10.5/
root@ubuntu:/home/cby/Tensorflow/Tensorflow/hdf5-1.10.5# ./configure --prefix=/usr/include/hdf5
root@ubuntu:/home/cby/Tensorflow/Tensorflow/hdf5-1.10.5# make
root@ubuntu:/home/cby/Tensorflow/Tensorflow/hdf5-1.10.5# make install
配置环境变量及软连接
export CPATH="/usr/include/hdf5/include/:/usr/include/hdf5/lib/"
root@ubuntu:/home/cby/Tensorflow/Tensorflow/hdf5-1.10.5# ln -s /usr/include/hdf5/lib/libhdf5.so /usr/lib/libhdf5.so
root@ubuntu:/home/cby/Tensorflow/Tensorflow/hdf5-1.10.5# ln -s /usr/include/hdf5/lib/libhdf5_hl.so /usr/lib/libhdf5_hl.so
安装whl包
root@ubuntu:/home/cby/Tensorflow/Tensorflow# pip3.7.5 install ./Cython-0.29.21-py2.py3-none-any.whl
root@ubuntu:/home/cby/Tensorflow/Tensorflow# pip3.7.5 install ./h5py-2.10.0-cp37-cp37m-linux_aarch64.whl
root@ubuntu:/home/cby/Tensorflow/Tensorflow# pip3.7.5 install ./grpcio-1.30.0.tar.gz
root@ubuntu:/home/cby/Tensorflow/Tensorflow# pip3.7.5 install ./gast-0.2.2.tar.gz
root@ubuntu:/home/cby/Tensorflow/Tensorflow# pip3.7.5 install ./opt_einsum-3.3.0-py3-none-any.whl
root@ubuntu:/home/cby/Tensorflow/Tensorflow# pip3.7.5 install ./Keras_Applications-1.0.8-py3-none-any.whl
root@ubuntu:/home/cby/Tensorflow/Tensorflow# pip3.7.5 install ./Keras_Preprocessing-1.1.2-py2.py3-none-any.whl
root@ubuntu:/home/cby/Tensorflow/Tensorflow# pip3.7.5 install ./astor-0.8.1-py2.py3-none-any.whl
root@ubuntu:/home/cby/Tensorflow/Tensorflow# pip3.7.5 install ./typing_extensions-3.7.4.3-py3-none-any.whl
root@ubuntu:/home/cby/Tensorflow/Tensorflow# pip3.7.5 install ./zipp-3.4.1-py3-none-any.whl
root@ubuntu:/home/cby/Tensorflow/Tensorflow# pip3.7.5 install ./importlib_metadata-3.10.1-py3-none-any.whl
root@ubuntu:/home/cby/Tensorflow/Tensorflow# pip3.7.5 install ./Markdown-3.2.2-py3-none-any.whl
root@ubuntu:/home/cby/Tensorflow/Tensorflow# pip3.7.5 install ./tensorboard-1.15.0-py3-none-any.whl
root@ubuntu:/home/cby/Tensorflow/Tensorflow# pip3.7.5 install ./wrapt-1.12.1.tar.gz
root@ubuntu:/home/cby/Tensorflow/Tensorflow# pip3.7.5 install ./tensorflow_estimator-1.15.1-py2.py3-none-any.whl
root@ubuntu:/home/cby/Tensorflow/Tensorflow# pip3.7.5 install ./termcolor-1.1.0.tar.gz
root@ubuntu:/home/cby/Tensorflow/Tensorflow# pip3.7.5 install ./tensorflow-1.15.0-cp37-cp37m-linux_aarch64.whl
注意：必须依次安装
安装Pytorch
root@ubuntu:/home/cby/pytorch/Pytorch# pip3.7.5 install ./apex-0.1+ascend-cp37-cp37m-linux_aarch64.whl
root@ubuntu:/home/cby/pytorch/Pytorch# pip3.7.5 install ./torch-1.5.0+ascend.post2-cp37-cp37m-linux_aarch64.whl
root@ubuntu:/home/cby/pytorch/Pytorch# pip3.7.5 install ./future-0.18.2.tar.gz
该文章所配套的软件包关注微信公众号回复 ai 即可获取所需要的所有软件包
