Comparing RDKit on a M1 MacBook Pro Max with Intel MacBookPro (2016)

https://www.rdkit.org

The RDKit is an open source toolkit for cheminformatics, 2D and 3D molecular operations, descriptor generation for machine learning, etc. There’s also a molecular database cartridge for PostgreSQL and cheminformatics nodes for KNIME (distributed from the KNIME community site: https://www.knime.org/rdkit)

The RDKit core algorithms and data structures are written in C++. Wrappers are provided to use the toolkit from either Python (2.x and 3.x), Java, or C#.

RDKIt was installed on both machines using miniconda

There are a standard set of benchmarks that run with the RDKit in order to detect systematic performance improvements or regressions. Those are here:

https://github.com/rdkit/rdkit/blob/master/Regress/Scripts/new_timings.py

https://github.com/rdkit/rdkit/blob/master/Regress/Scripts/new_timings.py

The associated data files are in the folder

https://github.com/rdkit/rdkit/blob/master/Regress/Data

The scripts run through a variety of cheminformatics operations

The command used was

The script was run 3 times and the fastest times shown below.

The Results

Intel timeM1 max timeM2 Air timeM2 Mac Studio Ultra
INFO: mols from smiles 
Results1: 11.08 seconds, 50000 passed, 0 failed 
INFO: Writing: Canonical SMILES 
Results2: 4.99 seconds 
INFO: mols from sdf 
Results1: 3.96 seconds, 10000 passed, 0 failed 
INFO: patterns from smiles 
Results3: 0.04 seconds, 823 passed, 0 failed 
INFO: Matching1: HasSubstructMatch 
Results4: 22.94 seconds 
INFO: Matching2: GetSubstructMatches 
Results5: 22.94 seconds 
INFO: reading SMARTS 
Results6: 0.01 seconds for 428 patterns 
INFO: Matching3: HasSubstructMatch 
Results7: 90.56 seconds 
INFO: Matching4: GetSubstructMatches 
Results8: 85.82 seconds 
INFO: Writing: Mol blocks 
Results10: 15.20 seconds 
INFO: BRICS decomposition 
Results11: 27.28 seconds 
INFO: Generate 2D coords 
Results12: 9.39 seconds 
INFO: Generate topological fingerprints 
Results16: 78.53 seconds 
INFO: Generate morgan fingerprints 
Results16: 3.31 second		INFO: mols from smiles 
Results1: 4.11 seconds, 50000 passed, 0 failed 
INFO: Writing: Canonical SMILES 
Results2: 2.16 seconds 
INFO: mols from sdf 
Results1: 1.60 seconds, 10000 passed, 0 failed 
INFO: patterns from smiles 
Results3: 0.02 seconds, 823 passed, 0 failed 
INFO: Matching1: HasSubstructMatch 
Results4: 13.55 seconds 
INFO: Matching2: GetSubstructMatches 
Results5: 13.67 seconds 
INFO: reading SMARTS 
Results6: 0.01 seconds for 428 patterns 
INFO: Matching3: HasSubstructMatch 
Results7: 56.01 seconds 
INFO: Matching4: GetSubstructMatches 
Results8: 50.21 seconds 
INFO: Writing: Mol blocks 
Results10: 7.35 seconds 
INFO: BRICS decomposition 
Results11: 13.51 seconds 
INFO: Generate 2D coords 
Results12: 4.88 seconds 
INFO: Generate topological fingerprints 
Results16: 51.80 seconds 
INFO: Generate morgan fingerprints 
Results16: 1.61 seconds		INFO: mols from smiles 
Results1: 3.9 seconds, 50000 passed, 0 failed 
INFO: Writing: Canonical SMILES 
Results2: 2.36 seconds 
INFO: mols from sdf 
Results1: 1.64 seconds, 10000 passed, 0 failed 
INFO: patterns from smiles 
Results3: 0.02 seconds, 823 passed, 0 failed 
INFO: Matching1: HasSubstructMatch 
Results4: 13.13 seconds 
INFO: Matching2: GetSubstructMatches 
Results5: 12.96 seconds 
INFO: reading SMARTS 
Results6: 0.01 seconds for 428 patterns 
INFO: Matching3: HasSubstructMatch 
Results7: 50.4seconds 
INFO: Matching4: GetSubstructMatches 
Results8: 45.82 seconds 
INFO: Writing: Mol blocks 
Results10: 6.83 seconds 
INFO: BRICS decomposition 
Results11: 17.51 seconds 
INFO: Generate 2D coords 
Results12: 4.65 seconds 
INFO: Generate topological fingerprints 
Results16: 46.80 seconds 
INFO: Generate morgan fingerprints 
Results16: 1.36 seconds		INFO: mols from smiles 
Results1: 3.3 seconds, 50000 passed, 0 failed 
INFO: Writing: Canonical SMILES 
Results2: 1.9 seconds 
INFO: mols from sdf 
Results1: 1.3 seconds, 10000 passed, 0 failed 
INFO: patterns from smiles 
Results3: 0.01 seconds, 823 passed, 0 failed 
INFO: Matching1: HasSubstructMatch 
Results4: 13.13 seconds 
INFO: Matching2: GetSubstructMatches 
Results5: 13.0 seconds 
INFO: reading SMARTS 
Results6: 0.01 seconds for 428 patterns 
INFO: Matching3: HasSubstructMatch 
Results7: 52 seconds 
INFO: Matching4: GetSubstructMatches 
Results8: 47 seconds 
INFO: Writing: Mol blocks 
Results10: 6.8 seconds 
INFO: BRICS decomposition 
Results11: 17.51 seconds 
INFO: Generate 2D coords 
Results12: 4.3 seconds 
INFO: Generate topological fingerprints 
Results16: 38.9 seconds 
INFO: Generate morgan fingerprints 
Results16: 1.2 seconds

These all measure single core performance.

Pharmacelera have created an open-source python script for conformation generation genConf.py. This script generates conformations plus a number of filters to generate a diverse selection of reasonable conformations. This is very typical workflow and as such is good measure of likely performance benefit.

genConf.py script workflow generated by Pharmacelera

The script is available for download here

Link to conformer script 3.0: https://pharmacelera.com/rdkit-conformer-generation-script-python-3/ 
Link to conformer script 2.7: https://pharmacelera.com/blog/scripts/rdkit-conformation-generation-script/

Again I used a selection of 1000 random structures from ChEMBL.

The Intel MacBook Pro took 4 hours 43 mins 
The MacBook Pro M1 max took 2 hours 46 mins.

List of tools tested https://macinchem.co.uk/software-reviews/cheminformatics-and-compchem-on-apple-silicon/

Last update 4 July 2023