This is the mail archive of the gcc-patches@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: limit code expnasion on ifcvt

From: Jan Hubicka <jh at suse dot cz>
To: Richard Henderson <rth at redhat dot com>, Jan Hubicka <jh at suse dot cz>,gcc-patches at gcc dot gnu dot org
Date: Thu, 21 Nov 2002 23:37:28 +0100
Subject: Re: limit code expnasion on ifcvt
References: <20021117134150.GH10371@kam.mff.cuni.cz> <20021121213510.GA12336@redhat.com>

> On Sun, Nov 17, 2002 at 02:41:50PM +0100, Jan Hubicka wrote:
> > 	* ifcvt.c (noce_try_store_flag_constants, noce_try_store_flag_inc,
> > 	noce_try_cmove_arith): Use maybe_hot_bb_p to limit code expansion.
> 
> Have you measured code shrinkage and performance impact for
> anything besides x86?
No.
There is small difference even on x86 (about 1Kb overall on SPEC
binaries).  The main code bloat appears to come from BRANCH_COST usage
outside ifcvt.c
Overall I would like to have code expnading transformations
conditionalized by the profile predicates.  What architecture would you
like to test this patch on?

Honza
> 
> 
> r~

Follow-Ups:
- Re: limit code expnasion on ifcvt
  - From: Richard Henderson

References:
- limit code expnasion on ifcvt
  - From: Jan Hubicka
- Re: limit code expnasion on ifcvt
  - From: Richard Henderson

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