From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-patches-return-133105-listarch-gcc-patches=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 25946 invoked by alias); 8 Nov 2004 20:04:58 -0000
Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Archive: <http://gcc.gnu.org/ml/gcc-patches/>
List-Post: <mailto:gcc-patches@gcc.gnu.org>
List-Help: <mailto:gcc-patches-help@gcc.gnu.org>
Sender: gcc-patches-owner@gcc.gnu.org
Received: (qmail 25891 invoked from network); 8 Nov 2004 20:04:51 -0000
Received: from unknown (HELO atrey.karlin.mff.cuni.cz) (195.113.31.123)
  by sourceware.org with SMTP; 8 Nov 2004 20:04:51 -0000
Received: by atrey.karlin.mff.cuni.cz (Postfix, from userid 29025)
	id B7DEB4B42F9; Mon,  8 Nov 2004 21:04:50 +0100 (CET)
Date: Mon, 08 Nov 2004 20:08:00 -0000
From: Zdenek Dvorak <rakdver@atrey.karlin.mff.cuni.cz>
To: Richard Henderson <rth@redhat.com>, gcc-patches@gcc.gnu.org
Subject: Re: [patch] Speed up ivopts
Message-ID: <20041108200450.GA18515@atrey.karlin.mff.cuni.cz>
References: <20041106205353.GA5351@atrey.karlin.mff.cuni.cz> <20041108191039.GF3278@redhat.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20041108191039.GF3278@redhat.com>
User-Agent: Mutt/1.5.6i
X-SW-Source: 2004-11/txt/msg00638.txt.bz2

Hello,

> On Sat, Nov 06, 2004 at 09:53:53PM +0100, Zdenek Dvorak wrote:
> > ! static const unsigned nzb_tab[] =
> 
> This is not ok.  At minimum you can put this out of line.  But as for
> how to use __builtin_ctz, see the use of __builtin_clz in toplev.h.
>
> > + /* Advance to the next nonzero bit in the intersection of
> > +    complemented bitmaps.  We will have already advanced past the just
> > +    iterated bit.  */
> > + 
> > + static inline bool
> > + bmp_iter_and_compl (bitmap_iterator *bi, unsigned *bit_no)
> > + {
> > +   if (bi->bits)
> > +     {
> > +       bmp_iter_advance (bi, bit_no);
> > +       return true;
> > +     }
> > +   return bmp_iter_and_compl_1 (bi, bit_no);
> 
> With the bits-in-word part factored out and inlined, does it make
> sense to move bmp_iter_and_compl_1 (etc) out of line?

I do not think so.  Unless these functions are fully inlined, sra won't
decompose the iterators and we are going to lose a lot on compile time
performace (or do you mean something else by "putting out of line"?

Zdenek