Skip to content

Latest commit

 

History

History
77 lines (61 loc) · 2.61 KB

README.md

File metadata and controls

77 lines (61 loc) · 2.61 KB
Raw

Plots for scientific computing


1. Introduction

This repository provides 2D plot templates for output from scientific computing codes focusing on parallel benchmarking data.


2. How to use this program?

Assign the directory where you want the plots to be stored in the source file plot.py.

plot_directory = 'plots'    # Directory to store plots

Comparison plots

Comparison plots can be generated from output data of various simulations. The inputs to the code are as follows:

comparison_directory   = 'results/comparison'         # Root directory for comparative input data
comparison_case_name   = ['mach_0p2', 'mach_0p4']     # Folder names for comparison cases
comparison_plot_labels = ['Mach 0.2', 'Mach 0.4']     # Plot labels for comparison cases

The plot_xy() function can then be used to generate the figure by reading appropriate column data from text files.

Parallel benchmarking plots

Parallel benchmarking plots can also be generated by giving the following inputs.

benchmark_directory = 'results/scaling'     # Root directory for benchmark input data
benchmark_partitions = [4, 8, 16, 32]       # Partition list and folder names for benchmark cases used in ascending order (starting from reference case).
benchmark_plot_labels = ['4 cores', '8 cores', '16 cores', '32 cores']  # Plot labels for benchmark cases

The total simulation run times can be plotted by using the function total_runTimes() from parallel benchmark data set. The minimum of these times for each partitioned run can be plotted as well with plot_bar_chart_minimum() function.

The most important data to plot for parallel benchmarking is the speedup and efficiency which can be obtained with plot_parallel_speedup() and plot_parallel_efficiency() functions.


3. How to run this program?

This program is written using python version 3 and it can be obtained by cloning this repository.

git clone https://github.com/jibranhaider/plotting

The program can then be executed by running the python script inside the cloned directory.

python3 plot.py

4. Author

This code is written by Jibran Haider.


5. License

This program is released under the MIT License. More details can be found in the LICENSE file.