This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: if-conversion a performance bottleneck
- To: Brad Lucier <lucier at math dot purdue dot edu>
- Subject: Re: if-conversion a performance bottleneck
- From: Michael Matz <matzmich at cs dot tu-berlin dot de>
- Date: Thu, 4 May 2000 20:45:08 +0200 (MET DST)
- cc: Richard Henderson <rth at cygnus dot com>, gcc at gcc dot gnu dot org
Hi,
On Wed, 3 May 2000, Brad Lucier wrote:
> Your changes to flow.c have cut the number of calls to
> sbitmap_intersection_of_succs from 40604 to 24225, so they are definitely
> worthwhile.
Yes, that was the purpose of the patch ;)
Anyway, I saw some issues of that patch with one of your other test files,
where the runtime exploded, and on my HDD at home I have another version
which even more reduces the time. If I find time, I'll rewrite the sbitmap
routines for large homogeneous bitmaps, clean up all that stuff and see if
anything comes out of it ;)
Btw. do you have also slightly smaller test cases (a quarter or so)? At
home I only have a slow machine and waiting 10 minutes everytime to see,
if one change was worthwhile is, erhh... slow.
> rest of compilation : 1.07 ( 0%) usr 0.00 ( 0%) sys 1.07 ( 0%) wall
> TOTAL : 226.10 1.27 227.32
God! For me that was ca. 1110 seconds ;) (Now I'm down to 580)
Ciao,
Michael.