Re: [PATCH] Fix PR26135, store copyprop being not effective

[Richard Guenther <rguenther at suse dot de>]

On Wed, 8 Feb 2006, Richard Guenther wrote:

> On Tue, 7 Feb 2006, Richard Guenther wrote:
> 
> > This adds the missing parts to the copyprop implementation to do store
> > copyprop.
> > 
> > Bootstrapped and tested on x86_64-unknown-linux-gnu, plain and with
> > -fno-tree-dominator-opts (as most store copyprop opportunities are
> > already handled by DOM).
> 
> And here is an alternative patch, killing current store copyprop and
> replacing it with a more powerful one walking virtual use-def chains
> (but at the same time making the stores propagation barriers due to
> limitation to single-outputs of the propagator).
> 
> As this is (with this patch) only done for store-copyprop and 
> store-copyprop runs very late (we could move it before PRE I suppose),
> I'm currently re-doing SPEC testing.

Results attached.  For eon store_copyprop runs too late it seems 
(compared to the earlier results).

Richard.

Attachment: CINT2000.038.asc
Description: Text document

Attachment: CFP2000.038.asc
Description: Text document