Conformer performance - Githubissues

openforcefield / openff-benchmark

Comparison benchmarks between public force fields and Open Force Field Initiative force fields

MIT License

10 stars 2 forks source link

Conformer performance #25

Open jthorton opened 3 years ago

jthorton commented 3 years ago

Description

This PR wraps the conformer generation function with a process pool to help speed up the step. By default, the pool will use all available processors but I have also exposed the argument to the CLI.

The time taken to generate conformers for the ~200 jacs set of ligands is reduced from ~6mins to 30s.

Questions

[ ] are there any drawbacks to allowing this?

Status

[ ] Ready to go

jthorton commented 3 years ago

It looks like there might be some broken tests for the compute side of things but the conformer tests are still working.

j-wags commented 3 years ago

Cool. Thanks, Josh. I'm behind on my testing for conformer gen, so I may add some more tests to this branch to help me sleep at night. I'll go ahead and merge when this is ready!

jthorton commented 3 years ago

No problem, I also need to explicitly code up a slightly different path when the number of processors is 1 as I got some strange behaviour in testing I should be able to do this tomorrow if you can hold off on the merge.

j-wags commented 3 years ago

Sounds good. I'll add the tests to master directly, so you'll get them when you update this branch.