Skip to content

Optimize runtime async OSR resumption performance #120865

@jakobbotsch

Description

@jakobbotsch

Resuming inside OSR methods is a slow process. It requires first calling the tier0 version which then calls into the runtime to perform an (unwinder-based) OSR transition.

We should not need a call into the runtime to transition to the OSR method from the tier0 method. Instead we can do a non-local jump from the tier 0 version when we realize we have an OSR method continuation.

Metadata

Metadata

Assignees

Labels

area-CodeGen-coreclrCLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMIruntime-asynctenet-performancePerformance related issue

Type

No type

Projects

No projects

Milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions