Adam C. Powell, IV (adam.powell@nist.gov)
Wed, 25 Nov 1998 11:13:53 -0500
Greetings and welcome to axp-linux-perf, the mailing list for high performance
on Linux-Alpha.
This list is intended as a discussion forum on the subject of maximizing the
performance of software running under Linux on alpha processors. There are
already several mailing lists and at least one newsgroup devoted to Alpha
Linux, but they are not intended for detailed discussions on performance
tuning, and this list will help filter this type of content out of the
background of hardware/kernel/installation messages that seem to comprise a lot
of the content of those forums (fora?).
Because the platform has such strong floating point performance and less
commercial software (browsers, office suites, etc) than Linux/Alpha, my guess
is that the list content will lean in scientific/engineering discussions;
because of the nature of Linux, there will probably be a focus on open-source
software; but the list is not intended to be exclusive to either of these.
Also, many subscribers showed an interest in Extreme Linux/Beowulf clusters, so
we'll probably have some discussions on that topic. Benchmarking is another
subject that might come up here, but let's try to avoid the kind of childish
competitiveness that characterizes much of this discussion on the mainstream
lists. ("Oh yeah? You think your code is hot? Well MY code runs TWENTY TIMES
FASTER on an 8-way AlphaServer 8400 than yours on a dual-PII!")
My dream is that the discussions on this list will give rise to some very high
performance open-source libraries and software tools (in the spirit of Wesner
and Goto's libffm and Goto's assembler BLAS), which will make it much easier
and more economical to take full advantage of the awesome power of the alpha
processor in many fields of research. But you each have your own dreams, and
this list will hopefully help you realize some of them.
Within 24 hours of the announcement, there were about 45 requests to be added
to the list, including all four "celebrities" mentioned (Goto, Wesner,
Hausmann, Payne), several DEC/Compaq employees, and one of the Sandia cplant
team members (see the alphalinux.org high-performance page), so we have quite a
team of experts here.
Now for some administrative details... As you all know, the server currently
hosting this list is kinda weak. I have had four offers/suggestions for
majordmo-style hosting, which would get rid of the lag times in
additions/deletions, restrict posting to list members, and make life easier for
me; I'll let the list know if such a change will take place. Until then, try
not to post the email list name on the web (though I guess it's inevitable in
the archive), but direct people to me to be added. This is because it is
impossible with this list server to restrict posting to list members, and so
anyone with a search engine can find the list name and spam it. I've seen this
happen to other lists on this server, and would rather it didn't happen to this
one.
Rich Payne is putting up an archive of the list at alphalinux.org.
That's all from me, let the discussions begin!
Zeen,
-Adam `Cold Fusion' Powell, IV http://www.ctcms.nist.gov/~powell/ ____
USDoC, National Institute of Standards & Technology (NIST) |\ ||< |
Center for Theoretical and Computational Materials Science | \||_> |
This archive was generated by hypermail 2.0b3 on Wed Nov 25 1998 - 03:36:41 EST