SergeyShch01 wrote:

@snehasish,

>Have you tested this end to end on a benchmark suite like SPEC?

Yes. I tested the complete workflow on SPEC CPU 2017. The binaries optimized 
with SPE-generated profiles showed performance comparable to those optimized 
with BRBE-generated profiles.

> I think we should document this support in 
> https://clang.llvm.org/docs/UsersManual.html#using-sampling-profilers and 
> include the settings users should use just like existing perf invocation.

Done. I added the Arm SPE recording command, requirements, and the 
corresponding llvm-profgen invocation to the manual.

https://github.com/llvm/llvm-project/pull/223237
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to