Skip to content

Segregate the register iterations based on intType/floatType #83109

@kunalspathak

Description

@kunalspathak

We noticed in #79544 (comment) and lately in #82731 that we should not iterate over all the register sets (specially do not iterate over float/SIMD if we know that there are no variables in the method). This will improve the TP performance.

Today, at multiple hot places, we iterate over all the registers.

for (regNumber reg = REG_FIRST; reg < ACTUAL_REG_COUNT; reg = REG_NEXT(reg))
{
 ...
}

Some other TP improvement potential:

Metadata

Metadata

Assignees

Labels

area-CodeGen-coreclrCLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI

Type

No type

Projects

No projects

Milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions