Python is probably the most widely used programming language in Cheminformatics at the moment and several readers have asked me to run a couple of benchmark tests. I’ve already tested a Jupyter notebookand that worked fine, the RDKit comparison also includes a fair amount of python code. Fortunately there are a few benchmarks that are well documented here https://benchmarksgame-team.pages.debian.net/benchmarksgame/index.html.
I used a couple of the benchmarks using the python implementation
Mandelbrot
Plot the Mandelbrot set [-1.5-i,0.5+i] on an N-by-N bitmap. Write output byte-by-byte in portable bitmap format.
Command used
|
1 |
time python -OO mandelbrot.py 16000 |
N-body
Model the orbits of Jovian planets, using the same simple symplectic-integrator.
Command used
|
1 2 |
time python -OO nbody.py 50000000 |
Binary-Trees
Command used
|
1 2 |
time python -OO binarytrees.py 21 |
The timings are shown in the table below.
| Test | Intel time | M1 max time | M2 Air time | M2 Mac Studio |
|---|---|---|---|---|
| Mandelbrot | 2 min 44s | 34s | 54s | 15 s |
| N-body | 7 min 17s | 3 min 57s | 4 min 21s | 4 min 2s |
| BinaryTrees | 46 sec | 12 secs | 16 secs | 6 secs |
List of tools tested https://macinchem.co.uk/software-reviews/cheminformatics-and-compchem-on-apple-silicon/
Last update 4 July 2023