[geeks] Re: HP's squandering

[Joshua D Boyd]

On Mon, Jan 13, 2003 at 01:41:39PM -0500, Dave McGuire wrote:
> On Monday, January 13, 2003, at 11:38 AM, Jochen Kunz wrote:
> >[discontinuing HP 3000]
> >>>>So WHY THE FUCK are they discontinuing them??
> >>>Because it is an obsolete, non-industry standard platform. ;-(
> >>   Oh yeah, right.
> >>   And the Itanium is enjoying so much more popularity, with such a
> >>huge installed base!
> >Yes, of course. Every big vendor has Itanium machines. Only HP has
> >HP3000.
> 
>   Some big vendors are *making* Itanium machines.  Nobody is buying 
> them, or has bought them.  There is a huge installed base of HP3000 
> machines.

The University of Amsterdam bought an Itanium "cluster" (is it really a
cluster if it only has two nodes? Anyway, both nodes are quad proc).
OSC has an itanium cluster of 72 dual proc nodes.  The NCSA (they are
the super computer center in Chicago, right?) has a 160 node (dual proc
nodes) cluster.  University of Manchester has a Itanium cluster, but
they are unspecific about how many nodes.  A university in JP claims
that Itaniums are 3x faster than 933mhz P3s, which we already know are
better at many types of FP than P4s.  Business users are somewhat more
difficult to find. 

So, it seems that Itaniums are getting out, just not in the server or
workstation market.  Which makes sense, since the sorts of people who
would accept such a generic workstation as the first Itaniums seem to be
the sort who would say P4 is god and pinch pennies on every single
part.  And Itaniums seem to fail to meet almost every part of the server
market at this point.  They aren't as cheap or dense as celerons on a
blade, and they don't have applications like SAP, DB/2, or Oracle on
them.

Perhaps that will change.  I'm kinda hopefull at the moment since I
think that this is a big improvement over the x86 line, and I like the
idea of such explicit paralelization on a single processor.

-- 
Joshua D. Boyd