RE: [PATCH, Loop optimizer]: Add logic to disable certain loop optimizations on pre-/post-loops

["Fang, Changpeng" <Changpeng dot Fang at amd dot com>]

Thanks, Jack:

I am going to update the patch against the current trunk!

Thanks,

Changpeng

________________________________________
From: Jack Howarth [howarth@bromo.med.uc.edu]
Sent: Friday, December 17, 2010 10:55 AM
To: Fang, Changpeng
Cc: Zdenek Dvorak; Richard Guenther; Xinliang David Li; gcc-patches@gcc.gnu.org
Subject: Re: [PATCH, Loop optimizer]: Add logic to disable certain loop optimizations on pre-/post-loops

On Fri, Dec 17, 2010 at 10:13:24AM -0600, Fang, Changpeng wrote:
>
> Hi, Jack:
>
> Have you tested my original patch (I remembered you bootstrapped it) in this thread?
> That is with the same logic which only applies to prefetch, unswitch and unrolling?
>
> It helps a lot if you have data about that?
>
> Thanks,
>
> Changpeng
>
>

Changpeng,
   I noticed that neither of your patches cleanly apply to recent gcc trunk due to
the context change in gcc/basic-block.h around...

  /* Set on blocks that cannot be threaded through.
     Only used in cfgcleanup.c.  */
  BB_NONTHREADABLE_BLOCK = 1 << 11,

  /* Set on blocks that were modified in some way.  This bit is set in
     df_set_bb_dirty, but not cleared by df_analyze, so it can be used
     to test whether a block has been modified prior to a df_analyze
     call.  */
  BB_MODIFIED = 1 << 12
};

You might want to rebenchmark pb05 against current gcc trunk to make sure
that r167779 didn't impact the results.
         Jack