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]

Re: scheduler speedup take 2


> > +   if (CHECK_DEAD_NOTES)
> > +     {
> > +       /* Remove all death notes from the subroutine.  */
> > +       for (rgn = 0; rgn < nr_regions; rgn++)
> 
> No, you cannot check death notes for multi-block regions.
> Interblock scheduling does not have the property of preserving
> the number of deaths.
Hmm, I was trying to come with example where the number of deaths changes
and failed.  Also I did bootstrap on sparc and it seemed to work.  Can you
englighten me by some example?

Updated patch is comming shortly.

Honza
> 
> 
> r~


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