LLVM vs stock GCC

Henri Gomez henri.gomez at gmail.com
Mon Apr 18 23:09:48 PDT 2011


>> The core functionality that Henri's test are evaluating are effectively complete. I would speculate that the minor drop in performance is likely the combination of forcing stack pointer alignment to 16-byte boundaries, and using -Os instead of -O3, as opposed to any systemic issue with LLVM.
>>
>> It would be interesting to find the combination of options that cause xalan, lusearch, and eclipse to fall behind the other OpenJDK variants. If there is an issue with using the LLVM backend, a bug should be filed against it ASAP (of course comparing -Os to -Os or -O3 to -O3, and ensuring all other factors are equal).
>
> I could redo the test of course. Did there is a way to change globally
> LLVM CC+C++ flags ?

I was thinking about CFLAGS and CXXFLAGS, if they are provided to make
command line.

BTW, I checked my open-bsd port, and I'm using gcc-4.0 (and not
gcc-4.2). I'll redo a new build to check if the perf diff couldn't
came from here.


More information about the macosx-port-dev mailing list