This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Performance testing of c-decl rewrite
- From: Daniel Jacobowitz <drow at false dot org>
- To: Zack Weinberg <zack at codesourcery dot com>
- Cc: gcc at gcc dot gnu dot org, Geoff Keating <geoffk at geoffk dot org>
- Date: Sun, 4 Apr 2004 11:21:32 -0400
- Subject: Re: Performance testing of c-decl rewrite
- References: <87vfkgjha1.fsf@egil.codesourcery.com>
On Sun, Apr 04, 2004 at 12:09:42AM -0800, Zack Weinberg wrote:
>
> I built GCC snapshots immediately before and after the patch and
> compiled X Windows (being a large collection of mostly-C source code
> that I happened to have lying around) with both, using oprofile to
> collect performance data. The machine was otherwise completely idle.
>
> Before (numbers are CPU_CLK_UNHALTED counts - an arbitrary scale,
> smaller is better)
>
> 23633380 59.4095 cc1
> 4338756 10.9068 vmlinux
> 3817502 9.5964 libc-2.3.2.so
> 3028414 7.6128 perl
> 1074030 2.6999 as
>
> After:
>
> 23767768 57.9481 cc1
> 5383616 13.1258 vmlinux
> 3839303 9.3606 libc-2.3.2.so
> 3058201 7.4562 perl
> 1072668 2.6153 as
>
> So, at first glance, this does look notably slower. Breaking it down
> a bit more, before:
Comparing different runs is one of oprofile's weakest points in my
experience. I recommend trying something old-fashioned like "time"
also.
When profilng GDB I found a lot of times that oprofile results, and
especially CLK_CPU_UNHALTED, would often scale by a constant factor
between runs. The only thing that was reliably interesting was the
changing order of functions in the profile.
--
Daniel Jacobowitz
MontaVista Software Debian GNU/Linux Developer