This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]
Other format:	[Raw text]

Re: compile time regression

From: Paolo Bonzini <paolo dot bonzini at lu dot unisi dot ch>
To: Diego Novillo <dnovillo at redhat dot com>
Cc: gcc at gcc dot gnu dot org, Richard Guenther <rguenther at suse dot de>
Date: Tue, 22 Aug 2006 21:00:08 +0200
Subject: Re: compile time regression
References: <20060822135628.ee69bd5f.bkoz@redhat.com> <44EB4EAF.50309@redhat.com>

Diego Novillo wrote:

Benjamin Kosnik wrote on 08/22/06 07:56:

Hey y'all. I'm just getting back from vacation and as I re-build my
testing baselines, I've noticed a huge compilation time regression.
This happened sometime post Aug 1, 2006. Anybody else notice?

Yes, something did happen after Aug11.  The SPEC testers show a 2-3%
bootstrap time increase from Aug11 to Aug13.

I don't see anything but this:

2006-08-11 Richard Guenther <rguenther@suse.de>

        PR middle-end/28651
        * simplify-rtx.c (simplify_const_relational_operation):
        Simplify A CMP B to A - B CMP 0 only for EQ and NE comparison
        codes.

but a 2-3% increase seems a lot! If it was this patch, it should be possible to find the place in GCC (or libiberty or libcpp) that is being badly pessimized.

I'm starting a bootstrap timing comparison to find if this is a red herring. I'll get the results next morning.

Paolo

Follow-Ups:
- Re: compile time regression
  - From: Paolo Bonzini

References:
- compile time regression
  - From: Benjamin Kosnik
- Re: compile time regression
  - From: Diego Novillo

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]