Skip to content

Benchmarks: use context pooling (almost) everywhere #26083

Open
@roji

Description

@roji

All our benchmarks should be using context pooling, to avoid the overhead of context creation interfering with results. We should still keep one benchmark to continue measuring pooling vs. non-pooling perf.

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions