Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Inverstigate excessive memory usage when running micro benchmarks. #5001

Open
@DimitrisJim

Description

@DimitrisJim

Summary

Some benchmarks result in excessive memory consumption, my system with approximately 15gb available for evaluating them cannot handle it. Specifically, mircro benchmarks that result in wild memory usage are currently:

  • call_simple.py
  • call_kwargs.py
  • define_function.py
  • define_class.py

These typically perform many allocations in a loop so this could just be chalked up to the missing GC.

Note: other benchmarks similarly perform relatively poorly on this front, i.e construct_object.py takes approximately 3-4gb on my machine. The cause is probably the same here.

Metadata

Metadata

Assignees

No one assigned

    Labels

    A-vmArea: virtual machine

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions