Skip to content

Optimize runtime countl_zero #1948

@andreaslyn

Description

@andreaslyn

Branch miss predictions in countl_zero have significant performance impact now that the interpreter is using it for the CLZ instruction. Change the runtime countl_zero function to a branchless implementation.

Metadata

Metadata

Labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions