Intelligent computing system 1 environment construction

1. Intelligent computing system

Intelligent computing system is a book , Written by chenyunji , This book uses the front end python tensorflow Declarative programming and frameworks , middle Deep learning processor , Bottom The Cambrian MLU270, A book that covers machine learning everything , It's no exaggeration to say , Figure out this book , Machine learning is half a master

2. Introduction to the experimental environment

This book contains many experiments , such as TensorFlow , such as bangc How to achieve CPU and MLU Heterogeneous execution, etc ; The test environment can be applied in the Cambrian computing platform (CPU/ MLU270 a sheet / Storage 1T), Storage should be permanent , CPU/MLU You can apply for half a month at most at one time ; That is, you can apply for a ECS belt through the platform MLU270, The effective time is 15 God , This is a container , To give you one IP:Port, You can go through xshell, VScode Etc ssh Connection to use , Used to Alibaba cloud , Tencent cloud Or have been using remote server Students who do experiments should be able to get started quickly

3. Experiment preparation

1. Win10 Just computers
2. VScode Installing a plug-in remote-ssh

4. Test environment

Go to the Cambrian development platform to apply for hardware

Storage request 1T No matter what , I applied here 20G, Enough can be used.
Cloud server , choice

5. adopt vscode Inside remote-ssh Connect to server

Starting environment , The code is contained in /opt in AICSE-demo-student.tar.gz Cambricon-Test

AICSE-demo-student.tar.gz Is all the environment and code
Cambricon-Test It's an experiment data And experiments model

cp AICSE-demo-student.tar.gz /home/zjq/
cd /home/zjq/ && tar -vxf AICSE-demo-student.tar.gz
#  Later, the code will be changed in /home/zjq Inside ,  Because this folder is attached to a separate storage volume ,  Even if the ECS expires ,  This can also bake data 

The following articles cover various bangc operator Experiment and tf Style shift of experiment