Quick manual for users of SDEval

Author: Albert Heinle <albert.heinle@googlemail.com>

Abstract

This document describes how to use the programs in this project for your everyday life.

For clarification on some denotation used in this document, consider the definitions

What is SDEval

SDEval is a collection of tools that can be used for benchmarking several computer algebra systems on different computation problems given problem instances from the Symbolic Data project.

It can be used to produce executable code for computer algebra systems, even a collection of them, and to run and time measuring them on different machines.

How can I use SDEval

You can use SDEval to create a task that you want to run on a certain machine. Furthermore, the execution of the task is also done by SDEval.

Creating a Task

The most handy way to do so is to use the graphical user interface. Fot that, just execute

$ python create_tasks_gui.py

on your machine. The first window that pops up shows the collection of currently provided computation problems.

_images/window1.png

Selection of the provided computation problems.

You choose one, and click next. Then a window similar to the following appears.

_images/window2.png

Selection of problem instances from Symbolic-Data

There, you can choose certain problem instances you want to have calculations to be run at. For a preview, simply double click on an entry in the left list box. If you want to add it, mark it and click the add button. Once you have selected all problem instances you want to use, click a last time on the Next-Button.

Then a window appears, where you can select different computer algebra systems.

_images/window3.png

Selection of computer algebra systems and setting the commands

Additionally, you have to add the commands to execute the selected computer algebra systems on your target machine. Also for a time command ist asked, which is on most UNIX-Like machines simply time. After giving your task a name, you can click on the Create-Button, and then you will be asked where to save the taskfolder. Select a location. After that, the program creates a taskfolder in the specified path.

Note

All those steps can also be done using the command line. For that, simply execute

$ python create_tasks.py

Executing a Task

Once you have created a task as described above, move to the taskfolder. There, you can execute the task by running the command

$ python runTasks.py

Note

You can also add time constraints and memory usage constraints. You can find out how to do that by typing

$ python runTasks.py --help

After you do that, your computer seems to be working. But you do not know what it is doing. The answer is: It is executing all files on the corresponding computer algebra systems, which might take a while.

To see the proceedings of the computations, go into the subfolder results, which is created after the first running, and select the folder with the name representing the current timestamp. In there, you find a HTML-file called proceedings.html. Open it, and see which computations are waiting, running and which are completed.

For the completed, you can open resultedtimings.html, where you can see how long the calculation took.

That is it. Have fun using it!!!

Table Of Contents

Previous topic

SymbolicData

Next topic

Definitions of terms used in the SDEval Project

This Page