Closed ChristopherEeles closed 1 year ago
Ditto for map_df
. There are probably more places that I haven't encountered yet.
I see the environmental variable inside of detect_cores
now and will use that to configure my parallelization.
Thanks, we will start working on this enhancement during the current sprint.
@ChristopherEeles the solution with BiocParallel
has been already implemented and merged into master. We'd be happy for your feedback.
Hi gDR team,
Lack of ability for users to control parallelization inside of
create_SE
is very nearly crashing my computer. I run out of RAM and end up swapping.This is happening with a relatively small dataset (6x6 drug combo, ~500 experiments), so I imagine it will only get worse as I try to incorporate larger ones.
Have you considered use of
BiocParallel
so users can configure their parallelization back-end themselves? It has the added benefit that you can pass in a "SerialParam" object for debugging, which enables stepping through execution.At minimum you need a way to cap the number of threads used.
I have worked around this for now, but it will likely cause trouble for other users.
Best, Chris