This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
First bug found :-)
- To: <gcc at gcc dot gnu dot org>
- Subject: First bug found :-)
- From: "M. Edward Borasky" <znmeb at aracnet dot com>
- Date: Tue, 22 May 2001 07:19:39 -0700
Well, I've found my first bug. I need to spend a little more time isolating
it, but briefly, when compiling the ATLAS library for the Athlon, one of the
routines appears to be getting numerical errors with the 20010514 gcc
snapshot. The same routine works fine with the default "gcc" on Red Hat 7.1,
which I believe is 2.96.81. By the way, the ATLAS main web page
(http://netlib.org/atlas) appears to be down at the moment, but the mirror
at Bell Labs is working (http://netlib.bell-labs.com/netlib/atlas/).
The timings for the routines that do work appear to be pretty much the same
with both compilers, given the normal amount of variation in benchmarks. I
need to clarify a few details with the Netlib folks; for example, they're
defining a cache size with -D that doesn't agree with another setting for
the same parameter and I need to find out whether it uses the -D setting or
not.
--
M. Edward (Ed) Borasky, Chief Scientist, Borasky Research
http://www.borasky-research.net http://www.aracnet.com/~znmeb
mailto:znmeb@borasky-research.com mailto:znmeb@aracnet.com
If there's nothing to astrology, how come so many famous men were born on
holidays?