------- Comment #16 from law at redhat dot com 2005-11-10 18:26 -------
Subject: Re: [4.1 Regression] Slowdown of the
bresenham line drawing by roughly 20%
On Wed, 2005-11-02 at 14:32 +0000, hubicka at ucw dot cz wrote:
> Hmm, perhaps restricting the reassociation + simplification into case
> where it kills last use of the intermediate result?
That's the only thing I had come up with too (see comment #14). I don't
particularly like that idea, probably because it's a hack.
But I like to be practical, and reality is the simple reassociation done
by DOM doesn't make nearly as much sense when the parent statement
doesn't die as a result of the optimization.
Jeff
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23181
