This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: optimization/4714: Haney's real matrix performance regression
- From: Paolo Carlini <pcarlini at unitus dot it>
- To: Daniel Berlin <dan at dberlin dot org>
- Cc: rth at gcc dot gnu dot org, CGLeggett at lbl dot gov, gcc-bugs at gcc dot gnu dot org, gcc-prs at gcc dot gnu dot org, nobody at gcc dot gnu dot org, pete at ltoi dot iap dot physik dot tu-darmstadt dot de, gcc-gnats at gcc dot gnu dot org
- Date: Wed, 03 Apr 2002 19:11:48 +0200
- Subject: Re: optimization/4714: Haney's real matrix performance regression
- References: <Pine.LNX.4.44.0204031152540.10772-100000@dberlin.org>
Daniel Berlin wrote:
>General induction variable.
>
>for (i = 0; i < 50; i ++)
>{
> a = i * 5;
>}
>
>i is a biv (basic induction variable).
>a is a giv, since it's a linear function of a biv.
>
Thanks Daniel for the clear explanation and example.
Indeed, the recent improvements to C++ alias analysis (coauthored by
yourself) is responsible for a much better behaviour of 3.2 on this
testcase. I'm not sure if it's already optimal, however.
Ciao, Paolo.