chronograph

This project times and analyzes the divsufsort() function in the parallel-divsufsort library. It runs the function with a range of thread counts, and then creates a graph of the results using matplotlib.

download

To download and configure this project, execute the following commands. Note that you will need bash and git installed to start.

git clone https://github.com/tgsachse/chronograph.git
cd chronograph
sudo bash chronograph.sh --install

sample texts

To download sample text files from the Internet, run this command:

bash chronograph.sh --download-samples

The texts come from Pizza&Chili and are saved into a folder called samples.

usage

This project analyzes the divsufsort() function and graphs the results separately. To analyze the function and produce a results data file, execute the following command:

bash chronograph.sh --analyze <file> <max_threads> <max_runs>

• <file> : The path to the text file you'd like to analyze.
• <max_threads> : The (integer) maximum thread count you'd like to analyze the function over. The function will be tested for [1, max_threads] threads.
• <max_runs> : The (integer) maximum number of times the function should be called for each thread count. The timings of these calls will be averaged out, so a higher max_runs value will increase accuracy but will also take longer to execute.

To create a sleek line graph with your new data, execute the following command:

bash chronograph.sh --graph <file>

• <file> : The path to the data file you'd like to graph.

example

The following example takes the input file my_stuff.txt, and tests divsufsort() on a range of [1, 32] threads. For each thread count, the function is called 10 times. The final results are saved in a data file named my_stuff.out:

bash chronograph.sh --analyze my_stuff.txt 32 10

Next, we graph the resulting data file like so:

bash chronograph.sh --graph my_stuff.out

This produces a new graph image named my_stuff.png.