This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: loop optimizations, mark 9
- To: law at cygnus dot com, Richard Henderson <rth at cygnus dot com>
- Subject: Re: loop optimizations, mark 9
- From: Richard Henderson <rth at cygnus dot com>
- Date: Thu, 21 May 1998 20:17:07 -0700
- Cc: egcs at cygnus dot com
- References: <19980519011620.A20650@dot.cygnus.com> <10945.895796228@hurl.cygnus.com>
- Reply-To: Richard Henderson <rth at cygnus dot com>
> The x86 doesn't bootstrap with these changes. Unfortunately c-torture
> doesn't show any regressions with these changes, so we don't have a
> simple testcase handy :(
That's a shame. Something to look into.
> I thought we had addressed the issue which originally led to your
> disabling of the 2nd loop optimization pass? Did I mis-understand
> something the last time we discussed these patches?
No, we'd never addressed that.
> Performance on spec92 on the x86 was kind-of hit or miss.
With spec95 on alpha, not all even ran successfully, though I don't
recall any performance regressions. Though that test run also included
the new gcse bits, so it's not as clear-cut as it might have been.
r~