[Beowulf] Three notes from ISC 2006

[Vincent Diepeveen]

Not at all good marketing that third remark.

Because if there was really something interesting to report,
then it would already have been reported by the *official* marketing 
department.
Good news travels fast and usually companies don't care much for NDA's there
and tell good news their biggest competitor/criticaster as first... :)

Note i liked seeing the first 2 as they are relevant for those who want good 
performance nodes in their beowulfs
and it involved a new processor just released this monday.

Of course this should not get default posting when a chip clocks another 
100Mhz higher or when some new
release of a compiler happens, as we know how completely obsolete by now 
specfp2000 is. Note it's hard
to expect future spec's to get much better, as the limit was 150MB for 
spec2004 and you must sign a form for spec
that THEY can sell YOUR code, and your competitor for a couple of thousands 
can get YOUR code in THEIR hands,
so any serious code never makes it to spec of course; my best versions i 
simply can't sell to spec for just 4000 dollar,
further their requirements on determinism are note serious, you can simply 
do floating point calculations with 32 bits 'doubles'
(if that would be faster) and suddenly look faster at benchmarks, as the 
'outcome' of the calculation wasn't influenced,
which the compiler team knew in advance as they have been testing for years 
with the outcome needed of course.

Basically having a big L2 or L3 matters a lot there and quite little in real 
world where the matrix calculations
performed don't fit in your L2 :)

----- Original Message ----- 
From: "Joe Landman" <landman at scalableinformatics.com>
To: <beowulf at beowulf.org>
Sent: Wednesday, June 28, 2006 3:45 PM
Subject: Re: [Beowulf] Three notes from ISC 2006


>
>
> Greg Lindahl wrote:
>> On Wed, Jun 28, 2006 at 08:28:06AM -0400, Mark Hahn wrote:
>>
>>> the "I know something that I can't tell" bit was childish though ;)
>>
>> Indeed, it was. I plead jet-lag.
>
> No.  It was good marketing.  Anyone on the list not at least a little
> curious what it is that Greg can't talk about?   (not shilling for
> PathScale, we are not a reseller for them, yet).
>
>
> -- 
> Joseph Landman, Ph.D
> Founder and CEO
> Scalable Informatics LLC,
> email: landman at scalableinformatics.com
> web  : http://www.scalableinformatics.com
> phone: +1 734 786 8423
> fax  : +1 734 786 8452 or +1 866 888 3112
> cell : +1 734 612 4615
> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) visit 
> http://www.beowulf.org/mailman/listinfo/beowulf
>