This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: gcc300 benchmarks slower than gcc295.3?
- To: ronis at onsager dot chem dot mcgill dot ca, gcc at gcc dot gnu dot org
- Subject: Re: gcc300 benchmarks slower than gcc295.3?
- From: "RenE J.V. Bertin" <rjvbertin at hotmail dot com>
- Date: Tue, 17 Jul 2001 19:33:51 +0200
>
>While 3.0 shouldn't give worse results than 2.95.3, note that stack
>alignment only works as well as your libraries and startup modules are
>aligned. Most installed libc's aren't, and hence, no matter how well
>3.0 maintains alignment, chances are your code isn't aligned.
>
>To fix this you have to rebuild libc with your preferred alignment
And find nothing works after reboot? :)
>options (although not with -malign-double!), and AFAIK, 3.0 still
>won't build glibc 2.2.3. (Anybody have an update on this?)
I am not sure whether this is the reason for the phenomenon I observe.
Firstly, as far as I have verified, the dhrystone benchmark programmes I
used (attached) don't do calls to library functions in the test routines
(strcmp should be inlined). Secondly, why would this give a different
penalty to gcc300 vs gc295.3?? I'd expect them to be penalised equally -
unless the default for e.g. -mpreferred-stack-boundary is different!
Another possibility is that gcc3 needs a more recent libc/kernel combination
(1.2.10/2.2.13-7mdk) than I am using. I'd hope not - I'd hate to have to
update my OS more than once every 2 years! :)
Attachments:
dry.c : integer dhrystones benchmark
fldry.c: floating point version of dry.c (I compiled with -Dextended=double)
_________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.
dry.c
fldry.c