This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: [OT] GCC vs Intel C++ compiler benchmark - don't forget ia64
- From: Andreas Jaeger <aj at suse dot de>
- To: "Scott Robert Ladd" <scott at coyotegulch dot com>
- Cc: <gcc at gcc dot gnu dot org>
- Date: Mon, 28 Jan 2002 07:21:45 +0100
- Subject: Re: [OT] GCC vs Intel C++ compiler benchmark - don't forget ia64
- References: <LHENLEBEOHJKKKONFNIPKEAMCOAA.scott@coyotegulch.com>
"Scott Robert Ladd" <scott@coyotegulch.com> writes:
>> this translates to time, effort and availability, all
>> three which intel has a lot of.
>
> Then perhaps it is in gcc's interest to focus on processors
> other than Intel's. Has anyone approached AMD about assistance
> im making sure that gcc produces the best possible code on
> Athlon/Duron processors?
There's an Athlon optimization guide - like there're Intel
optimizations guide out. I've contacts to AMD folks that can give
sometimes additional insight into optimization issues.
The SPEC testing of GCC I do (http://www.suse.de/~aj/SPEC) uses AMD
Athlon systems.
There're different optimizations for let's say Athlons:
- scheduling etc. This has been done long ago (included in GCC 3.0
for Athlon) and smaller adjustments are done over the time.
- generic optimizations that need platform specific support,
e.g. recently prefetch
- generic optimizations that help all architectures but might have a
different impact on different architectures
Generic optimizations are most needed and I expect that those are
difficult to convince most companies to support since they help
everybody.
Andreas
--
Andreas Jaeger
SuSE Labs aj@suse.de
private aj@arthur.inka.de
http://www.suse.de/~aj