The source code and dockerfile for the GSW2024 AI Lab.
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
This repo is archived. You can view files and clone it, but cannot push or open issues/pull-requests.
 
 
 
 
 
 

2.2 KiB

Thread local memory example
===========================

The GLPK library, when compiled with default options, uses a separate environment
for each thread that is executed. So each thread is isolated. The only exeption
is error handling. An error in any of the library functions will not only
terminate the active thread but the complete process.

This can be circumvented by defining an error handling routine with
glp_error_hook(). This directory contains an example demonstrating running a
multithreaded application with error handling.


The example code
----------------

The program multiseed solves a MathProg model multiple times in separate parallel
threads. Each threads uses a different seed for the MathProg pseudo random number
generator.

The MathProg model clustering.mod generates 50 randomly distributed "towns". Out
of the towns it selects 3 to be cluster centers and assign the other towns to the
clusters such that the sum of the population weighted euclidian distances between
towns and centers is minimized.

The solution is written to a Scalable Vector File which can be viewed with a web
browser.

For demonstration purposes at the end of every third thread the error handling
routine is triggered by calling glp_error(). This results in output like

18-00086 Model has been successfully processed
18-00087 Voluntarily throwing an error in multiseed.c at line 147
18-00088 Error detected in file multiseed.c at line 146
18-00089 Error caught

Terminal output is preceeded by numbers indicating the thread index and the
output line. You can pipe it through sort to get a better overiew, e.g.

multiseed clustering.mod 20 | sort


Building and running the example code
-------------------------------------

On Linux multiseed can be compiled with gcc by calling

make

The code can be executed with

make check

For compiling the example on 64bit Windows with Microsoft Visual Studio C++ run

Build_Multiseed.bat

You may have to adust the variable HOME in the batch file depending on the
installation path of Visual Studio.

You can run multiseed with

multiseed <filename> <count>

Where filename is the path to the model file and count is the number of parallel
threads.