Skip to content

Add MT benchmarking (aka investigate gil-release during extraction) #317

Description

@masklinn

Originally proposed by #316

Currently ua-parser-rs does not release the GIL during extraction, as noted in the above this was not a specific positive and supported decision, rather just keeping to the defaults (of sorts). So getting numbers would be a good idea to know:

  • the overhead of release/acquire the GIL on every extraction in ST
  • the same in MT
  • compare with free-threaded, ST and MT
  • while at it, check with subinterpreters (although that might complicate things a touch, the subinterpreter executor pickles parameters, so might require some sort of interpreter-local global and loading the library in each of them individually, TBD)
  • and subprocesses because why not

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