Skip to content

Parallel trace transformation #3096

@msoeken

Description

@msoeken

Is your feature request related to a problem? Please describe.
We could improve QRE runtime when expensive trace transforms are in place by applying trace transforms in parallel. This currently happens sequential in TraceQuery.enumerate even though resource estimation makes use of threading.

Describe the solution you'd like
Making this parallel in Python won't help much because most of the trace code is in rust and we would need pickling which requires a large amount of serialization (this may get easier if we re-factor the internal trace data type). A better way would be to move the enumeration code into Rust and apply multi-threading there.

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions