This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: comments on loop changes
- To: Richard Henderson <rth at cygnus dot com>
- Subject: Re: comments on loop changes
- From: Michael Hayes <m dot hayes at elec dot canterbury dot ac dot nz>
- Date: Thu, 04 Feb 1999 11:14:45 +1300 (NZDT)
- Cc: Michael Hayes <m dot hayes at elec dot canterbury dot ac dot nz>, Jorn Wolfgang Rennecke <amylaar at cygnus dot com>, egcs at cygnus dot com
- References: <19990131145827.A17537@cygnus.com><14004.58794.114591.617460@ongaonga.elec.canterbury.ac.nz><19990131171622.B17567@cygnus.com><14005.526.838270.939987@ongaonga.elec.canterbury.ac.nz><19990131214105.A17739@cygnus.com><14005.30140.529348.966766@ongaonga.elec.canterbury.ac.nz><19990201025925.A17750@cygnus.com>
Richard Henderson writes:
> I'll take a closer look at your reg+reg problem and see if I can
> replicate it on Sparc. There's got to be a way to convince the
> combiner to try reg+reg last.
I've found I can emulate the behaviour of the Sparc by setting
ADDRESS_COST to 1. I then get essentially the same pattern of GIVs
and the GIV combination works well.
So it appears that the problem I see is due to a subtle interaction
between CSE and GIV combination. Somehow we need the GIV combiner to
be a bit smarter but I'm unsure how to progress.
Michael.