Mark Hellegers wrote:

>
> Am I correct in that ? Do you mean that N2 is still on the roadmap ?
>
I won't mean anything unless I'm doing an NDA presentation, except what
I said: people are assuming that the R16K-800 is the last MIPS processor,
and the current roadmap does not indicate it is.

To get an idea about what architectural changes ther may be, etc. and
their estimated impact on performance, you'd have to get an NDA
presentation.

--
Alexis Cousein Senior Systems Engineer
alexis@sgi.com SGI/Silicon Graphics Brussels

Nobody Expects the Belgian Inquisition!

Alexis Cousein wrote:
>
> Mark Hellegers wrote:
>
> >
> > Am I correct in that ? Do you mean that N2 is still on the roadmap ?
> >
> I won't mean anything unless I'm doing an NDA presentation, except what
> I said: people are assuming that the R16K-800 is the last MIPS processor,
> and the current roadmap does not indicate it is.
>
> To get an idea about what architectural changes ther may be, etc. and
> their estimated impact on performance, you'd have to get an NDA
> presentation.

I don't know if you can answer this question, but I'll ask anyway: Is
there any indication as to when that information will be made publicly
available? I.e., how long before we can hear about this stuff outside
of an NDA presentation?

John-Paul Stewart wrote:

> I don't know if you can answer this question, but I'll ask anyway: Is
> there any indication as to when that information will be made publicly
> available? I.e., how long before we can hear about this stuff outside
> of an NDA presentation?

Well -- if SGI's historical behaviour holds, that will be the day the
processors are officially launched and appear on the SGI pricebook .


--
Alexis Cousein Senior Systems Engineer
alexis@sgi.com SGI/Silicon Graphics Brussels

Nobody Expects the Belgian Inquisition!

Alexis Cousein wrote:
> you'd have to get an NDA presentation.

And how does one go about that?

--
#include
Christopher Miller
cm007i@hotmail.com

<<>> wrote:

> Alexis Cousein wrote:
>
>> you'd have to get an NDA presentation.
>
>
> And how does one go about that?
>
Usually -- wave dollar bills in front of a salesperson. And
sign ("execute") a non-disclosure agreement.


--
Alexis Cousein Senior Systems Engineer
alexis@sgi.com SGI/Silicon Graphics Brussels

Nobody Expects the Belgian Inquisition!

Alexis Cousein wrote:
>
> John-Paul Stewart wrote:
>
> > I don't know if you can answer this question, but I'll ask anyway: Is
> > there any indication as to when that information will be made publicly
> > available? I.e., how long before we can hear about this stuff outside
> > of an NDA presentation?
>
> Well -- if SGI's historical behaviour holds, that will be the day the
> processors are officially launched and appear on the SGI pricebook .

Somehow that's exactly the kind of vague answer I was expecting.

Oh well. Never hurts to ask!

In article <4030B316.40708@brussels.sgi.com>,
Alexis Cousein wrote:

: Mark Hellegers wrote:
:
: >
: > Am I correct in that ? Do you mean that N2 is still on the roadmap ?
: >
: I won't mean anything unless I'm doing an NDA presentation, except what
: I said: people are assuming that the R16K-800 is the last MIPS processor,
: and the current roadmap does not indicate it is.
:
: To get an idea about what architectural changes ther may be, etc. and
: their estimated impact on performance, you'd have to get an NDA
: presentation.

Can we play the vague game?

The next mips chip will draw power:
a) more than current mips chips
b) about in line with current mips chips
c) less than current mips chips

The next mips chip will compare with the current itanic lineup:
a) favourably or b) unfavourably in a performance/watt context
c) favourably or d) unfavourably in a performance/dollar context
e) favourably or f) unfavourably in a performance/acerage context

The next mips chip will contain:
a) more of the same, slightly faster
b) more of the same, but twice as much of it
c) more of the same, plus something new and interesting glued on
d) mostly new and interesting stuff

The next mips chip will contain cache that is:
a) larger than the current mips chips
b) faster than the current mips chips
c) in more layers than the current mips chips

The next mips chip will be fabbed on a process that is:
a) wee
b) not so wee
c) FRIKKIN' HUGE

The next mips chip will assure SGI's:
a) doom
b) continued survival
c) world domination

The next mips chip will cheat on benchmarks:
a) a little, like a POWER chip
b) a lot, like an Itanic chip
c) completely, like a SPARC chip

The mips line of chips is doomed to failure like the:
a) Itanic
b) Alpha
c) PA-RISC
d) CLIPPER
e) M88000
f) Z8000


Ok, maybe the questions are a little unfair. I am looking forward to seeing
more new MIPS CPUs, and I do hope they're not a "same ****, different MHz"
snooze-fest.


Cheers - Tony 'Nicoya' Mantler

--
Tony 'Nicoya' Mantler -- Master of Code-fu -- nicoya@ubb.ca
-- http://nicoya.feline.pp.se/ -- http://www.ubb.ca/ --

Alexis Cousein schreef:
> Mark Hellegers wrote:
>>
>> Am I correct in that ? Do you mean that N2 is still on the roadmap ?
>>
> I won't mean anything unless I'm doing an NDA presentation, except what
> I said: people are assuming that the R16K-800 is the last MIPS processor,
> and the current roadmap does not indicate it is.
>
> To get an idea about what architectural changes ther may be, etc. and
> their estimated impact on performance, you'd have to get an NDA
> presentation.
>
I think all of us here don't need an NDA to read between the lines that there
will be speed bumps, r16k might reach 1 GHz, but the speed bumps will be slow
to arrive and certainly fail Moore's law.
Consequently the Irix installed base will decline. Perhaps this can be
compensated by growth of the Altix installed base. Looking at the current
value increase of SGI shares, quite some people seem to think so.

-- hans

On 2/17/2004 4:22 PM, HvdV wrote:
> Alexis Cousein schreef:
>
>> Mark Hellegers wrote:
>>
>>>
>>> Am I correct in that ? Do you mean that N2 is still on the roadmap ?
>>>
>> I won't mean anything unless I'm doing an NDA presentation, except what
>> I said: people are assuming that the R16K-800 is the last MIPS processor,
>> and the current roadmap does not indicate it is.
>>
>> To get an idea about what architectural changes ther may be, etc. and
>> their estimated impact on performance, you'd have to get an NDA
>> presentation.
>>
> I think all of us here don't need an NDA to read between the lines that
> there will be speed bumps, r16k might reach 1 GHz, but the speed bumps
> will be slow to arrive and certainly fail Moore's law.
> Consequently the Irix installed base will decline. Perhaps this can be
> compensated by growth of the Altix installed base. Looking at the
> current value increase of SGI shares, quite some people seem to think so.
>
> -- hans

My guess is that the increased stock price is more about debt
refinancing and corporate restructuring (e.g., selling Alias). I don't
know many details about Altix, but it seems that Intel is continuing to
torpedo Itanium, which doesn't bode well for Altix or SGI.

In article <1035g6fc2321eed@corp.supernews.com>,
Michael Rice wrote:
......
>My guess is that the increased stock price is more about debt
>refinancing and corporate restructuring (e.g., selling Alias).

Stock prices are their own non-linear non-deterministic system.
Unlike blood pressure, their variance is of little intrinsic
significance to the health of the corporation.

> I don't
>know many details about Altix, but it seems that Intel is continuing to
>torpedo Itanium, which doesn't bode well for Altix or SGI.

If you don't know many details, what do you base this on?

The very long-term design of the O3000/O300 line clearly allowed
for alternate cpus. The engineering in the Altix is of a piece with
the corresponding Origin. It only makes sense for SGI to adopt the
most cost-effective cpu available while leaveraging existing
engineering and OS development.

Now why in the world would Intel torpedo their only high-end
cpu? What they *have* found is that the intermediate cpus such
as the Opteron fill the 32/64 bit gap and allow penetration of the
workstation market and smooth customer migration. It is clear that
Intel didn't want to create a product that would cannabilize their
own high-end 32 bit cpu sales. Now, with competition, they need a
product in this niche.

--
Daniel Packman
NCAR/ACD
pack@ucar.edu

On 2/17/2004 9:06 PM, Daniel Packman wrote:
> In article <1035g6fc2321eed@corp.supernews.com>,
> Michael Rice wrote:
> .....
>
>>My guess is that the increased stock price is more about debt
>>refinancing and corporate restructuring (e.g., selling Alias).
>
>
> Stock prices are their own non-linear non-deterministic system.
> Unlike blood pressure, their variance is of little intrinsic
> significance to the health of the corporation.
>
>
>>I don't
>>know many details about Altix, but it seems that Intel is continuing to
>>torpedo Itanium, which doesn't bode well for Altix or SGI.
>
>
> If you don't know many details, what do you base this on?
>
> The very long-term design of the O3000/O300 line clearly allowed
> for alternate cpus. The engineering in the Altix is of a piece with
> the corresponding Origin. It only makes sense for SGI to adopt the
> most cost-effective cpu available while leaveraging existing
> engineering and OS development.

The basis for my statement is that effort spent by Intel on IA-64 is
effort that might have been spent on Itanium. I can't help but believe
this will result in even more delays in the IPF schedule - which is not
good at all for vendors with Itanium-based products. As IA-64 and
Operton peformance improves, the Itanium choice becomes even more
difficult for consumers. It just appears as though Itanium is being
pushed further into the HPC and high-end server niche.

I'm not a hardware engineer, but I'd think that changing to a different
CPU would take some serious engineering dollars - it's not like you can
just drop in an IA-64 where an Itanium used to be (is it?). Not to
mention the recurring costs (hardware production, software maintenance,
hardware/software support, etc.) in maintaining another product line.
SGI doesn't appear to bulging with excess engineering resources to
support these kinds of activities.

These are just my perceptions of recent activities of Intel, IBM, Sun,
and HP. Maybe the Opteron/IA-64 dust will settle and the primary focus
will shift back to Itanium - but it doesn't look like it will happen soon.

Michael Rice writes:

> The basis for my statement is that effort spent by Intel on IA-64 is
> effort that might have been spent on Itanium.

Please don't call the mooted 64-bit x86 stuff "IA-64" since IPF was
formerly called IA-64, and still is in many circles. IA-64 == IPF.

*p

> Please don't call the mooted 64-bit x86 stuff "IA-64" since IPF was
> formerly called IA-64, and still is in many circles. IA-64 == IPF.

That's right. It seems Intel calls the 64-bit x86 "IA32E"
See http://www.tecchannel.de/news/hardware/14533/

Michael Rice wrote:
>
> My guess is that the increased stock price is more about debt
> refinancing and corporate restructuring (e.g., selling Alias). I don't
> know many details about Altix, but it seems that Intel is continuing to
> torpedo Itanium, which doesn't bode well for Altix or SGI.

Oh, they're selling Alias? So that is likely to mean further decreasing support
for IRIX from them?

--
/ http://www.fishpool.com/~setok/

On 2/18/2004 2:07 AM, Per Ekman wrote:
> Michael Rice writes:
>
>
>>The basis for my statement is that effort spent by Intel on IA-64 is
>>effort that might have been spent on Itanium.
>
>
> Please don't call the mooted 64-bit x86 stuff "IA-64" since IPF was
> formerly called IA-64, and still is in many circles. IA-64 == IPF.
>
> *p

Oops, meant to call it x86-64.

pack@eos.ucar.edu (Daniel Packman) wrote in message news:...
> Now why in the world would Intel torpedo their only high-end
> cpu?

Because despite 10 odd years of development and marketing they don't
seem to be able to sell anything with the Itanic. Excepting some kits
from HP and a fair number of Altix sales from SGI.


Cheers,

Shawn

In article <33d2be31.0402182316.7d73bfa9@posting.google.com>,
Shawn Johnston wrote:
>pack@eos.ucar.edu (Daniel Packman) wrote in message
>news:...
>> Now why in the world would Intel torpedo their only high-end
>> cpu?

>Because despite 10 odd years of development and marketing they don't
>seem to be able to sell anything with the Itanic. Excepting some kits
>from HP and a fair number of Altix sales from SGI.

I'm sure it has proven to be a disappointment for many
parties involved, but with so much of the development cost
already expended, there is almost no turning back. The 64bit
extensions to the ia32 might not lead immediately to itanium
sales, but it provides a much-needed bridge for a software transition.

--
Daniel Packman
NCAR/ACD
pack@ucar.edu

Have to agree with that. Were thinking of porting our code from the 64bit
mips to ia64 enviroment now. It's just so much easier now we have ia64 and
more cost effective for the performance we will see.

*********************
Khalid Schofield
System Administrator / EM Technician
Dept. Of Materials
University Of Oxford
Parks Road
Oxford
OX1 3PH

Email: khalid.schofield@materials.ox.ac.uk
Tel: 01865 273785
Fax: 01865 283333
Web: http://www-em.materials.ox.ac.uk/peo...eld/index.html


On Thu, 19 Feb 2004, Daniel Packman wrote:

> In article <33d2be31.0402182316.7d73bfa9@posting.google.com>,
> Shawn Johnston wrote:
> >pack@eos.ucar.edu (Daniel Packman) wrote in message
> >news:...
> >> Now why in the world would Intel torpedo their only high-end
> >> cpu?
>
> >Because despite 10 odd years of development and marketing they don't
> >seem to be able to sell anything with the Itanic. Excepting some kits
> >from HP and a fair number of Altix sales from SGI.
>
> I'm sure it has proven to be a disappointment for many
> parties involved, but with so much of the development cost
> already expended, there is almost no turning back. The 64bit
> extensions to the ia32 might not lead immediately to itanium
> sales, but it provides a much-needed bridge for a software transition.
>
> --
> Daniel Packman
> NCAR/ACD
> pack@ucar.edu
>

Michael Rice wrote:
> The basis for my statement is that effort spent by Intel on IA-64 is

I believe Intel calls it IA-32e.

> effort that might have been spent on Itanium. I can't help but
> believe this will result in even more delays in the IPF schedule -
> which is not good at all for vendors with Itanium-based products. As
> IA-64 and Operton peformance improves, the Itanium choice becomes
> even more difficult for consumers. It just appears as though Itanium
> is being pushed further into the HPC and high-end server niche.

There are many entrenched places where IA-64 or IA64 is a synonym for
IPF (aka Itanium) (Linux is one of them), so it might be better to
call Intel's extentions to their IA-32 CPUs what I believe Intel calls
them - IA-32e.

rick jones
--
firebug n, the idiot who tosses a lit cigarette out his car window
these opinions are mine, all mine; HP might not want them anyway...
feel free to post, OR email to raj in cup.hp.com but NOT BOTH...