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: Another performance regression

From: Dale Johannesen <dalej at apple dot com>
To: Richard Henderson <rth at redhat dot com>
Cc: Dale Johannesen <dalej at apple dot com>, gcc-patches at gcc dot gnu dot org, gcc at gcc dot gnu dot org
Date: Fri, 27 Sep 2002 15:11:12 -0700
Subject: Re: Another performance regression

On Thursday, September 26, 2002, at 03:26 PM, Richard Henderson wrote:

On Thu, Sep 26, 2002 at 12:01:38PM -0700, Dale Johannesen wrote:

So should I fix this by making the unrolling code smarter, in effect
doing cse's job?

I think so, since the change to loop unrolling order was
also done to get around a performance regression.  :-/

It seems likely Roger Sayle's approach of running
gcse after loop opts would Just Work.  Is that going to go in?

Not for 3.3.

The following appears to do the job.  Would you see if this
causes any other problems?

Thanks!
Bootstrapped and tested on Darwin (C, C++, ObjC, f77), no regressions.

References:
- Re: Another performance regression
  - From: Richard Henderson

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