Archived from groups: comp.arch,comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

http://www.ictworld.co.za/Editoria [...] alID=20287

No sooner has HP stopped making workstations than SGI has stepped in to make
Itanium workstations. It looks like these workstations are a modification of
their existing Itanium servers, as they are supposedly capable of being put
together with anywhere between 2 and 512 processors! I doubt that anyone
would put 512 processors on a workstation, but it's intriguing to think
about it.

Yousuf Khan


--
Humans: contact me at ykhan at rogers dot com
Spambots: just reply to this email address ;-)

Archived from groups: comp.arch,comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

Yousuf Khan wrote:

> http://www.ictworld.co.za/Editoria [...] alID=20287
>
> No sooner has HP stopped making workstations than SGI has stepped in
> to make Itanium workstations.

Well, the Prism is not really comparable to what one usually understands as
a workstation. Unlike HPs cancelled desktop machines zx2000 and zx6000 the
SGI Prism is more a visualization supercomputer than a workstation, and thus
being the successor of the Onyx and not of SGIs desktop workstations (Fuel,
Tezro)...

Benjamin

Archived from groups: comp.arch,comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

"Yousuf Khan" <bbbl67@ezrs.com> writes:

> http://www.ictworld.co.za/Editoria [...] alID=20287
>
> No sooner has HP stopped making workstations than SGI has stepped in to make
> Itanium workstations. It looks like these workstations are a modification of
> their existing Itanium servers, as they are supposedly capable of being put
> together with anywhere between 2 and 512 processors! I doubt that anyone
> would put 512 processors on a workstation, but it's intriguing to think
> about it.

Calling an Altix with graphic cards a "workstation" is quite stretching
the term.

-Andi

Archived from groups: comp.arch,comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

Andi Kleen wrote:

> "Yousuf Khan" <bbbl67@ezrs.com> writes:
>
>> http://www.ictworld.co.za/Editoria [...] alID=20287
>>
>> No sooner has HP stopped making workstations than SGI has stepped in to
>> make Itanium workstations. It looks like these workstations are a
>> modification of their existing Itanium servers, as they are supposedly
>> capable of being put together with anywhere between 2 and 512 processors!
>> I doubt that anyone would put 512 processors on a workstation, but it's
>> intriguing to think about it.
>
> Calling an Altix with graphic cards a "workstation" is quite stretching
> the term.
>
> -Andi

Well, I don't see the word workstation in the marketing blurb (4-8u is
mentioned for the baby one), but having said that previous 'workstations'
such as fuel/octane were based on Origin systems so I am not sure that I
understand your objection to the concept?

Martyn


Martyn Foster
Performance Libraries
Silicon Graphics.

Archived from groups: comp.arch,comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

Yousuf Khan wrote:
> http://www.ictworld.co.za/Editoria [...] alID=20287
>
> No sooner has HP stopped making workstations than SGI has stepped in to make
> Itanium workstations. It looks like these workstations are a modification of
> their existing Itanium servers, as they are supposedly capable of being put
> together with anywhere between 2 and 512 processors! I doubt that anyone
> would put 512 processors on a workstation, but it's intriguing to think
> about it.

Odd, I encountered this story at Ars first. There the news was "up to
16 processors" and one day in the future up to 512 processors. This
article and the SGI tech sheet and press release don't seem to mention
any current barrier to 512P.

It's about time someone had some sense. All this time intel is trying
to force a chip to a market (Itanium 2->servers) instead of forcing a
market to move to their chip (workstation users->Itanium 2). Of course
they have been unsuccessful. Itanium 2 is not a server, it's a workstation.

Alex
--
My words are my own. They represent no other; they belong to no other.
Don't read anything into them or you may be required to compensate me
for violation of copyright. (I do not speak for my employer.)

Archived from groups: comp.arch,comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

On Tue, 12 Oct 2004 08:07:55 -0400, Alex Johnson wrote:

>
> Odd, I encountered this story at Ars first. There the news was "up to 16
> processors" and one day in the future up to 512 processors. This article
> and the SGI tech sheet and press release don't seem to mention any current
> barrier to 512P.

See http://www.sgi.com/products/visual [...] nfigs.html

And they'll problably build an even bigger one if you ask them nicely.

> been unsuccessful. Itanium 2 is not a server, it's a workstation.

Actually, it's a processor.

--
Ricardo

Archived from groups: comp.arch,comp.sys.ibm.pc.hardware.chips,comp.sys.intel (More info?)

On Tue, 12 Oct 2004 08:07:55 -0400, Alex Johnson <compuwiz@jhu.edu> wrote:

>Yousuf Khan wrote:
>> http://www.ictworld.co.za/Editoria [...] alID=20287
>>
>> No sooner has HP stopped making workstations than SGI has stepped in to make
>> Itanium workstations. It looks like these workstations are a modification of
>> their existing Itanium servers, as they are supposedly capable of being put
>> together with anywhere between 2 and 512 processors! I doubt that anyone
>> would put 512 processors on a workstation, but it's intriguing to think
>> about it.
>
>Odd, I encountered this story at Ars first. There the news was "up to
>16 processors" and one day in the future up to 512 processors. This
>article and the SGI tech sheet and press release don't seem to mention
>any current barrier to 512P.
>
>It's about time someone had some sense. All this time intel is trying
>to force a chip to a market (Itanium 2->servers) instead of forcing a
>market to move to their chip (workstation users->Itanium 2). Of course
>they have been unsuccessful. Itanium 2 is not a server, it's a workstation.

A lot of that (workstation) market relies on high-cost, complex,
independently developed, relatively low-volume software though and the
users can only use what's available. Many of the ISVs got burned with
Risc/6000s, Alphas and even, to a certain extent, Suns over the past
10years; even if the hardware system was essentially free to them, which it
usually wasn't, they are much less likely to take on the risks of
conversion/development and the maintenance burden a 2nd or 3rd time.

I believe that AMD64 really has closed any window of opportunity for
Itanium there. Intel's refusal to acknowledge the existence of Yamhill for
the past 2(?) years... followed by the eventual announcement of EM64T,
seems to indicate, to me anyway, that even they seem to have seen the
light. Any residual reluctance is, of course, understandable but the
"market" is speaking:
http://www.ptc.com/partners/hardwa [...] letter.htm (sorry if
you've already seen that) and putting it quite bluntly too.

Looking at the numbers, for what we paid IBM for monthly Risc/6000
maintenance charges a few years back, we can now get/build a workstation
class PC every 2 months or so. Hardware maintenance cost on a current
workstation would, of course, be much less than back then but I'd think it
would be a brave soul who would pass it up vs. how they would on a PC-based
system. Even without the initial cost, it just doesn't add up for Itanium.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??