Skip to content

Caching njit functions #78

@jensdebruijn

Description

@jensdebruijn

Kind of request

None

Enhancement Description

As you probably know, numba allows caching the njit compiled version of a function so that it doesn't have to be compiled next run.

I would like to ask whether it would be good to turn on this caching in pyflwdir. If you think so, I am happy to make a PR for this.

Use case

Models go brrr

GEB uses pyflwdir during model initalization. The repeated compilation is currently about 2/3 of model initializaton in small test regions that we use for debugging (the impact is of course much smaller in large regions).

Additional Context

No response

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions