[patch] restore bootstrap on ppc

Zdenek Dvorak rakdver@kam.mff.cuni.cz
Wed Sep 12 12:09:00 GMT 2007


Hello,

> > 	* fold-const.c (extract_muldiv_1): Do not simplify
> > 	var * c * c to var.
> 
> it looks like this is only performance related patch that might've
> caused today regression on gzip and gcc:
> http://www.suse.de/~gcctest/SPEC-britten/CINT/sandbox-britten/recent.html
> http://www.suse.de/~gcctest/SPEC-britten/CINT/sandbox-britten/200709111625.int/
> It seems quite serious, so hopefully there is way around (or it might be
> a noise even if there was no changes in setup, we should see next run
> relatively shortly)

if performance of gzip and gcc relied on optimizining var * 8 * 8 into
var, then I believe we are stuck -- I cannot bring this optimization
back, sorry :-)

Zdenek

> Honza
> > 
> > Index: fold-const.c
> > ===================================================================
> > *** fold-const.c	(revision 128320)
> > --- fold-const.c	(working copy)
> > *************** extract_muldiv_1 (tree t, tree c, enum t
> > *** 6273,6279 ****
> >   	  && ((code == MULT_EXPR && tcode == EXACT_DIV_EXPR)
> >   	      || (tcode == MULT_EXPR
> >   		  && code != TRUNC_MOD_EXPR && code != CEIL_MOD_EXPR
> > ! 		  && code != FLOOR_MOD_EXPR && code != ROUND_MOD_EXPR)))
> >   	{
> >   	  if (integer_zerop (const_binop (TRUNC_MOD_EXPR, op1, c, 0)))
> >   	    {
> > --- 6273,6280 ----
> >   	  && ((code == MULT_EXPR && tcode == EXACT_DIV_EXPR)
> >   	      || (tcode == MULT_EXPR
> >   		  && code != TRUNC_MOD_EXPR && code != CEIL_MOD_EXPR
> > ! 		  && code != FLOOR_MOD_EXPR && code != ROUND_MOD_EXPR
> > ! 		  && code != MULT_EXPR)))
> >   	{
> >   	  if (integer_zerop (const_binop (TRUNC_MOD_EXPR, op1, c, 0)))
> >   	    {



More information about the Gcc-patches mailing list