This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: big slowdown in egcs-1.1.2->gcc-2.95 on alpha
- To: Thomas E Deweese <deweese at kodak dot com>, Joern Rennecke <amylaar at cygnus dot co dot uk>, lucier at math dot purdue dot edu
- Subject: Re: big slowdown in egcs-1.1.2->gcc-2.95 on alpha
- From: Michael Meissner <meissner at cygnus dot com>
- Date: Fri, 5 Nov 1999 15:31:21 -0500
- Cc: gcc at gcc dot gnu dot org
- References: <14371.15034.186821.135519@gargle.gargle.HOWL>
On Fri, Nov 05, 1999 at 03:14:50PM -0500, Thomas E Deweese wrote:
>
> JR> Well, since it worked for one place that looks for set bits in
> JR> allocno sets, let's see if it works in the two other ones:
>
> I'm not 100% certain of the density of 1's in the bit fields
> but my impression is that they are fairly sparse (I'm guessing,
> especially in the 'conflicts' case). In which case I would suggest at
> least putting a check on any bits being set in 'word_' around the
> inner for loop.
If they are really large and sparse, bitmaps might be more effective than
sbitmaps (assuming you use the EXECUTE_ macro in bitmaps).
--
Michael Meissner, Cygnus Solutions
PMB 198, 174 Littleton Road #3, Westford, Massachusetts 01886
email: meissner@cygnus.com phone: 978-486-9304 fax: 978-692-4482