This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Widening vs PPro
- To: Richard Henderson <rth at cygnus dot com>
- Subject: Re: Widening vs PPro
- From: Jeffrey A Law <law at upchuck dot cygnus dot com>
- Date: Wed, 14 Apr 1999 01:31:57 -0600
- cc: Jan Hubicka <hubicka at atrey dot karlin dot mff dot cuni dot cz>, egcs-patches at egcs dot cygnus dot com
- Reply-To: law at cygnus dot com
In message <19990414001616.B868@cygnus.com>you write:
> On Wed, Apr 14, 1999 at 12:26:19AM -0600, Jeffrey A Law wrote:
> > The more I think about it, the more I suspect that defining PROMOTE_MODE
> > to force widening of all non-memory integer operations is probably the
> > best way to deal with partial registers stalls for the x86 variants where
> > it is an issue.
> >
> > It's probably worth an experiment for ...
>
> You'd think, but it didn't seem to be the case in practice. Even
> when I only promoted HImode operations.
>
> Some more extensive benchmarking is certainly in order, but...
Wow. It sounds like it actually made things worse. Real strange.
Oh well. Maybe someday we can actually dive into the results in more
detail :-)
jeff