This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug middle-end/16373] -O2 and -O3 compiler switches do not "omit-frame-pointers" as the docs. describe.
- From: "davejf at frontiernet dot net" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: 6 Jul 2004 01:15:33 -0000
- Subject: [Bug middle-end/16373] -O2 and -O3 compiler switches do not "omit-frame-pointers" as the docs. describe.
- References: <20040705185622.16373.davejf@frontiernet.net>
- Reply-to: gcc-bugzilla at gcc dot gnu dot org
------- Additional Comments From davejf at frontiernet dot net 2004-07-06 01:15 -------
I agree - what brought this up was I was benchmarking some code with a lot of
function calls and wondered why GCC was so much slower.. This switch always made
a pretty large difference - on the order of +40% or so for some.
Seems like a lot of debug related call overhead to add by default to optimized
"production" executables on what has to be the most widely GCC targeted CPU (x86).
Incidently, there seems to be many benchmark results floating around out there
that under-represent GCC performance because of this, not to mention code
running in the real-world. Given that other real popular compilers like MSVC++
and Intel omit frame pointers for simliar Ox switches, I think the currently
documented behaviour (frame pointers omitted for -Ox) would be well accepted.
Thanks,
Dave
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16373