Scheduling complete loop unrolling early and unconditional?

Richard Guenther rguenth@tat.physik.uni-tuebingen.de
Thu Dec 16 16:04:00 GMT 2004


On Thu, 16 Dec 2004, Diego Novillo wrote:

> Richard Guenther wrote:
>
> > Therefore, is it possible to schedule either complete unrolling
> > of loops very early (before sra), or to schedule another sra
> > run after loop optimizations (that's what I'll try at the moment,
> > but possibly mess everything up)?
> >
> Scheduling SRA after loop optimizations sounds better.  SRA should be
> able to run more than once.

Ok, I'm trying to bootstrap

Index: tree-optimize.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/tree-optimize.c,v
retrieving revision 2.65
diff -u -r2.65 tree-optimize.c
--- tree-optimize.c	30 Nov 2004 15:38:33 -0000	2.65
+++ tree-optimize.c	16 Dec 2004 16:02:34 -0000
@@ -381,6 +381,10 @@
   NEXT_PASS (pass_loop);
   NEXT_PASS (pass_dominator);
   NEXT_PASS (pass_redundant_phi);
+  NEXT_PASS (pass_sra);
+  NEXT_PASS (pass_rename_ssa_copies);
+  NEXT_PASS (pass_dominator);
+  NEXT_PASS (pass_redundant_phi);
   NEXT_PASS (pass_cd_dce);
   NEXT_PASS (pass_dse);
   NEXT_PASS (pass_forwprop);

possibly pass_rename_ssa_copies is required after sra?  And the
others may help?

Richard.



More information about the Gcc mailing list