COMPSs Tutorial applications

This folder contains tutorial applications developed for [py]COMPSs.

Its purpose is to share the tutorial applications among the COMPSs users community to ease the implementation of new applications and share them with the rest of users.

Tutorial Application's Folder Structure

Application are grouped by language (C, Java and Python). Each of them includes a set of applications used in tutorials following the next structure:

apps
  |- python
  |    |- app_1
  |    |    |- src                 Application source code
  |    |    |- README              Brief description of the application
  |- java
  |    |- app_1
  |    |    | ...
  |- c
       |- ...

Tutorial Hands-on Instructions and Commands

Setup Instructions

Update Tutorial apps:

$ rm -rf tutorial_apps 
$ git clone https://github.com/bsc-wdc/tutorial_apps.git

Install COMPSS-PyCOMPSS CLI:

$ sudo python3 -m pip install pycompss-cli

Python Hands-on

Initialize a PyCOMPSs docker environment.

$ pycompss init -n docker-tutorial docker -i compss/compss-tutorial:2.10

Start monitor.

$ pycompss monitor start

Start Jupyter Notebook.

$ pycompss jupyter

Once started, open a web browser with the address: http://localhost:8888 Follow the instructions of the Jupyter notebook, and at the end of the hands-on session, follow the instructions to clean the environment.

Stop the monitor.

$ pycompss monitor stop

Remove the PyCOMPSs docker environment.

$ pycompss environment remove docker-tutorial 

Java Hands-on

Compile the application.

$ cd /home/compss/tutorial_apps/java/wordcount
$ mvn clean install

Initialize a COMPSs docker environment.

$ compss init -n docker-tutorial docker -i compss/compss-tutorial:2.10

Follow the exercises to develop the wordcount application.

Run the application.

$ compss run --classpath=jar/wordcount.jar wordcount.uniqueFile.Wordcount data-set/file_small.txt 650

To enable an execution with monitoring:

Start the monitor.

$ compss monitor start

Open a browser in URL http://localhost:8080/compss-monitor and enter an log-in with an empty user.

Run the application with monitoring flags.

$ compss run -m --classpath=jar/wordcount.jar wordcount.uniqueFile.Wordcount data-set/file_long.txt 350000

Simulate the execution with different workers.

• To check the used resources run use the following commands:

$ compss exec cat /project.xml
$ compss exec cat /resources.xml

• Include two worker nodes.

$ compss components add worker 2

• Run again the application with monitoring flags

$ compss run -m --classpath=jar/wordcount.jar wordcount.uniqueFile.Wordcount data-set/file_long.txt 350000

Run the application with graph generation and generate the graph

$ compss run -g --classpath=jar/wordcount.jar wordcount.uniqueFile.Wordcount data-set/file_small.txt 650
$ compss gengraph .COMPSs/wordcount.uniqueFile.Wordcount_04/monitor/complete_graph.dot
$ evince .COMPSs/wordcount.uniqueFile.Wordcount_04/monitor/complete_graph.pdf

Run the application with debug information

$ compss run -d --classpath=jar/wordcount.jar wordcount.uniqueFile.Wordcount data-set/file_small.txt 650

Stop the monitor

$ compss monitor stop

Remove the docker environment

$ compss environment remove docker-tutorial

C++ exercise

Enter in the application folder and start the COMPSs docker environment

$ cd ~/tutorial_appss/c/matmul_object
$ compss init -n docker-tutorial docker -i compss/compss-tutorial:2.10

Build the application

$ compss exec compss_build_app Matmul

Run the application

$ compss run --appdir=/home/user/ master/Matmul 4 4 2.0

Remove the docker environment

$ compss environment remove docker-tutorial