From a2efc04da87a5b5c471bb423fb80111379adf66a Mon Sep 17 00:00:00 2001 From: Yanqingshang Date: Thu, 22 Oct 2020 10:15:14 +0800 Subject: [PATCH] modify readme file --- README.en.md | 57 +++++------------------------------------------ README.md | 62 +++++++++------------------------------------------- 2 files changed, 15 insertions(+), 104 deletions(-) diff --git a/README.en.md b/README.en.md index 816b01ea6..2e8287887 100755 --- a/README.en.md +++ b/README.en.md @@ -22,69 +22,22 @@ At the same time, the system meets the following requirements - [SWIG](http://www.swig.org/download.html) #### Download - +./build.sh ``` git clone https://gitee.com/ascend/tensorflow.git cd tensorflow ``` -#### Configure environment -```BASH -./configure +#### Execute script to generate installation package ``` - -By default, executing the above command will pop up the following interactive session window - -> Your session may be different. -```BASH -Please specify the location of python with available tensorflow v1.15.0 installed. [Default is /usr/bin/python] -(You can make this quiet by set env [ASCEND_TARGET_PYTHON_BIN_PATH]): +./build.sh ``` -At this point, you are required to enter the path of the python interpreter with Tensorflow v1.15.0 installed. If the default path is correct, press Enter, otherwise, please enter the correct python -Interpreter path. -> You can set ASCEND_TARGET_PYTHON_BIN_PATH Environment variable to suppress the pop-up of the interactive window, but still make sure that the path is valid, otherwise, you will still be asked to enter the correct path to the python interpreter. +After the script is successfully executed, a compressed file of tfadapter.tar will be generated in the output directory -After typing, it will take a few seconds to make sure your input is valid, and then the following interactive window will pop up -``` -Please specify the location of ascend. [Default is /usr/local/Ascend] -(You can make this quiet by set env [ASCEND_INSTALL_PATH]): -``` - -At this time, you are required to enter the installation path of the Ascend processor development kit. If the default path is correct, press Enter, otherwise, please enter the correct path to the Ascend processor development kit installation. - -> You can set ASCEND_INSTALL_PATH Environment variables to suppress the pop-up of interactive windows, but still make sure that the path is valid, otherwise, you will still be asked to enter the correct installation path of the Ascend processor development kit. - - -After typing, it will follow the interaction window below -``` -Please specify the location of swig. [Default is /usr/local/bin/swig] -(You can make this quiet by set env [SWIG_INSTALL_PATH]): -``` -At this time, you are required to enter the path of the SWIG executable file. If the default path is correct, press Enter, otherwise, please enter the correct path of the SWIG executable file. - -> You can suppress the interactive window pop-up by setting the SWIG_INSTALL_PATH environment variable, but make sure that the path is valid, otherwise, you will still be asked to enter the correct SWIG executable file path. - -After typing, wait for the configuration to complete. -#### Configure cmake - -``` -mkdir build -cd build -cmake .. -``` -#### Build -```BASH -make -j8 -``` - -After compilation, the installation package will be generated in -``` -./dist/python/dist/npu_bridge-1.15.0-py3-none-any.whl -``` #### Install - +Unzip the tfadapter.tar file to generate npu_bridge-1.15.0-py3-none-any.whl You can install the TF_Adapter plug-in using pip. ``` pip install ./dist/python/dist/npu_bridge-1.15.0-py3-none-any.whl diff --git a/README.md b/README.md index b5dbcd4b0..79ec66c9e 100755 --- a/README.md +++ b/README.md @@ -20,6 +20,9 @@ Tensorflow Adapter For Ascend（简称TF_Adapter）致力于将昇腾AI处理器 - GCC >= 7.3.0 - CMake >= 3.14.0 - [SWIG](http://www.swig.org/download.html) +- 配置昇腾处理器开发套件安装路径的环境变量D_LINK_PATH + export D_LINK_PATH=安装路径（例如：/usr/local/Ascend） + #### 下载源码 @@ -28,63 +31,18 @@ git clone https://gitee.com/ascend/tensorflow.git cd tensorflow ``` -#### 配置安装环境 -```BASH -./configure +#### 执行脚本生成安装包 ``` -默认情况下，执行上述命会弹出如下的交互式会话窗口 -> 您的会话可能有所不同。 -```BASH -Please specify the location of python with available tensorflow v1.15.0 installed. [Default is /usr/bin/python] -(You can make this quiet by set env [ASCEND_TARGET_PYTHON_BIN_PATH]): +./build.sh ``` -此时，要求您输入安装了 Tensorflow v1.15.0 版本的python解释器路径，如果默认路径是正确的，直接回车，否则请输入正确的 python -解释器路径。 -> 你可以通过设置 ASCEND_TARGET_PYTHON_BIN_PATH ->的环境变量，来抑制交互式窗口弹出，但是要确保路径是有效的，否则，仍然会要求您输入正确的 python 解释器路径。 -键入后，会耗费几秒钟以确保您的输入是有效的，接着，会弹出下面的交互式窗口 -``` -Please specify the location of ascend. [Default is /usr/local/Ascend] -(You can make this quiet by set env [ASCEND_INSTALL_PATH]): -``` -此时，要求您输入昇腾处理器开发套件的安装路径，如果默认路径是正确的，直接回车，否则请输入正确的昇腾处理器开发套件安装路径。 - -> 你可以通过设置ASCEND_INSTALL_PATH ->的环境变量，来抑制交互式窗口弹出，但是要确保路径是有效的，否则，仍然会要求您输入正确的昇腾处理器开发套件安装路径。 - -键入后，会接着弹出下面的交互式窗口 -``` -Please specify the location of swig. [Default is /usr/local/bin/swig] -(You can make this quiet by set env [SWIG_INSTALL_PATH]): -``` -此时，要求您输入SWIG可执行文件路径，如果默认路径是正确的，直接回车，否则请输入正确的SWIG可执行文件路径。 - -> 你可以通过设置SWIG_INSTALL_PATH ->的环境变量，来抑制交互式窗口弹出，但是要确保路径是有效的，否则，仍然会要求您输入正确的SWIG可执行文件路径。 +脚本执行成功后，会在output目录生成tfadapter.tar压缩文件 - -键入后，等待配置完成。 -#### 配置cmake -> 根据您的网络状况，可能需要数分钟来下载TF_Adapter的依赖项目以完成配置。 -``` -mkdir build -cd build -cmake .. -``` -#### 执行编译 -> 您应当根据实际编译环境，设置合适的并发编译数以提升编译速度。 -```BASH -make -j8 -``` -编译结束后，安装包会生成在 -``` -./dist/python/dist/npu_bridge-1.15.0-py3-none-any.whl -``` -#### 安装 -您可以使用 pip 安装 TF_Adapter 插件。 +#### 安装插件包 +解压tfadapter.tar文件，生成npu_bridge-1.15.0-py3-none-any.whl +使用 pip 安装 TF_Adapter 插件。 ``` -pip install ./dist/python/dist/npu_bridge-1.15.0-py3-none-any.whl +pip install npu_bridge-1.15.0-py3-none-any.whl ``` 需要注意的是， 您应当保证安装路径与您编译时指定的 python 解释器搜索路径是一致的。 -- Gitee