This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Improve unrolled size estimates


On Mon, May 11, 2009 at 10:50 AM, Richard Guenther <rguenther@suse.de> wrote:
> So you are basically doing constant propagation on the loop body with
> initial values from the entry edge. ?Why not use CCP for this if you
> go through all this hassle?

Is it really worth doing all the changes needed CCP for this simple
case, where the values on the entry edge always dominate the loop?  It
would only be a win to do CCP if there are expressions that fold to a
constant an are fed back via a back-edge to a PHI.  I don't know how
likely this is, but I'd dare a hand-wavey guess "not too common".

I'd just add a TODO about an opportunity for further optimizations
later on, but not hold this patch up until those CCP changes are
made...

Ciao!
Steven


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]