This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch] rtlopt branch merge part 6 - speed improvement of branch predictions
- From: Jan Hubicka <jh at suse dot cz>
- To: Richard Henderson <rth at redhat dot com>,Josef Zlomek <zlomj9am at artax dot karlin dot mff dot cuni dot cz>,gcc-patches at gcc dot gnu dot org
- Date: Fri, 7 Feb 2003 11:54:45 +0100
- Subject: Re: [patch] rtlopt branch merge part 6 - speed improvement of branch predictions
- References: <20030203101819.GA15255@artax.karlin.mff.cuni.cz> <20030207020048.GB28920@redhat.com>
> On Mon, Feb 03, 2003 at 11:18:19AM +0100, Josef Zlomek wrote:
> > * Makefile.in (sreal.o): Added.
> > (predict.o): Depends on sreal.h instead of real.h.
> > * sreal.c: New file.
> > * sreal.h: New file.
> > * predict.c: Use sreal.c instead of real.c.
>
> I wonder if the following would be better. Thoughts?
I was playing around with this, but this fails when the underlying FP
implementation changes - ie x87 math has double rounding even with
volitales everywhere and you get different results when you bootstrap
using x87 compiler compiler that uses SSE math.
Honza