JITs Benchmarks

Here I will try to explain how can you reproduce the results I got for ACO presentations 2022.

Requirements

• gcc
• bash
• julia (version >= 1.5)
• python (version >= 3.6)

One needs also the necessary packages for julia and python (TOML parser for C is provided by https://github.com/cktan/tomlc99 and already included in the project).

Please run the following on julia:

using Pkg
Pkg.add("StaticArrays")
Pkg.add("TOML")

Please run the following on your favourite terminal:

python -m pip install --user numpy numba tomli

Steps to get the results

1. Enter any of the folders.
2. Execute make.
3. Adapt template.toml to your needs. DO NOT CHANGE THE LINE WHICH SAYS "PATATA".
4. Execute bash run.sh.

Once this is done, the timings in seconds will be in:

• ccfile for gcc.
• cofile for gcc -O3.
• jlfile for julia.
• pyfile for python + numpy + numba.
• ppfile for python + numpy. These are tsv like files. The results I got in my computer are already included in the each of the folders

If you want some graphs, consider running (matplotlib and pandas is needed, you may install them by running python -m pip install --user pandas matplotlib).

python plots.py

in each folder.