This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: GCC Status (2004-08-09)
- From: Scott Robert Ladd <coyote at coyotegulch dot com>
- To: mitchell at mail dot codesourcery dot com
- Cc: gcc at gcc dot gnu dot org
- Date: Tue, 10 Aug 2004 00:09:17 -0400
- Subject: Re: GCC Status (2004-08-09)
- References: <20040809184502.31786.qmail@mail.codesourcery.com>
I've been away from GCC for a several weeks, on other business. I've
just returned, and will be applying various benchmarks in the next week
or so to GCC 3.5 and earlier.
mitchell@mail.codesourcery.com wrote:
In addition, it's not entirely clear how good the code generated from
GCC 3.5 is relative to the code generated from earlier releases. I've
heard everything from "much better" to "much worse"; my current guess
is "not too different".
I think the best characterization is "on average, not too different". In
some cases, 3.5 does better than 3.4 (and earlier); in other situations,
the opposite is true. The bottom line is that 3.5 does not (at this
time) provide siginificant improvements in generate code speed.
In any case, I'd like to start with (1). The vectorizer is unlikely
to do any harm; the most frequent complaint I've heard is that it
doesn't vectorize enough loops. And, vectorization -- even if
somewhat incomplete -- will be an exciting feature for GCC 3.5 that
will show what all the new optimization infrastructure can do.
I completely agree. Having a vectorizer (perfect or not) will eliminate
a common PR hit against GCC in certain circles.
--
Scott Robert Ladd
Coyote Gulch Productions (http://www.coyotegulch.com)
Software Invention for High-Performance Computing