Newisys AMD x86-64 servers

Are opterons the future of desktops or are they just a stop gap.

On Mon, 23 Jun 2003 13:34:33 -0400, Yousuf Khan wrote:


I am starting to see a definitive shift towards Linux solutions in
places where there was previously a Solaris-only shop. It's just an
observation I've noticed in the last couple of months.

Yousuf Khan


All of us in the Western World also see the solar eclipse ;-)

Seriously, a few engineering companies I worked in have
surrounded Sun servers with mostly Linux workstations where
in the past they would have bought all Sun workstations.

Rich Teer wrote:

On Mon, 23 Jun 2003, Yousuf Khan wrote:

Well, it's not just the number of processors that determine a server's
sophistication. It's also all of the advanced fault-tolerant hardware
onboard, and system management software available.

Still paying catch up in Intel land? Us SPARC users have had those
features on low end servers for years...

On the other hand Intel land has had full parity and EDC protected cache
hierarchies for a lot longer than Sun customers.

Q: If a Sun server crashes in a forest and no one not under NDA
observed it, did it really crash?


--
Paul W. DeMone The 801 experiment SPARCed an ARMs race of EPIC
Kanata, Ontario proportions to put more PRECISION and POWER into
architectures with MIPSed results but ALPHA's well
that ends well.




-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----== Over 80,000 Newsgroups - 16 Different Servers! =-----

"Bill Todd" wrote in message
...
The AMD Opterons on the other hand
are a completely different animal, they live for scaling

Up to 4 processors, certainly. Up to 8 processors, possibly. Beyond
that,
no one knows yet (at least no one who is talking).

Maybe the examples don't exist yet, but looking at the architecture it
certainly seems like it should not be a problem scaling beyond 8 processors
with external solutions.

-- each Opteron
controls its own memory directly and they talk to each other over
point-to-point links rather than shared buses.

Only up to 8 processors. What happens beyond that point has not yet been
disclosed, AFAIK.

At that point it would be a proprietary solution, something like Sun's
Fireplane interconnect. It doesn't have to Hypertransport through and
through. You could use HT between the processors within any single system
board, but you can connect each of the system boards with any other
interconnect you like. It's just a basic fact of life in NUMA architectures,
some connections are going to be slower than others.

Some people question how well Opteron will scale to 8 processors in a
glueless configuration. In particular, it uses a broadcast-invalidation
(rather than directory-based) cache-coherence mechanism that certainly
would
have difficulty scaling much beyond 8 processors and may scale less than
linearly in the 4 - 8 processor region. We'll presumably see soon whether
AMD just allowed that extension because it was easy to do so and because
it
scales usefully even if not very linearly.

I really don't see this as a problem. This would be a problem in a shared
memory architecture like Intel's, but each Opteron directly accesses its own
assigned memory, and when it comes time to access memory located on another
Opteron, each remote Opteron has to ask permission from the local Opteron's
controller.

There is already a Cray
supercomputer project for Sandia National Labs making use of 10,000
Opterons; that uses a special Cray proprietary crossbar connector
between
system boards, but the rest is just off-the-shelf.

My impression is that it's not ccNUMA, however. If so, while it may well
be
interesting for Cray, most operating systems would have no idea how to use
it.

There is also a project at some US university to use a 128 Opterons in a
system. I think it was the University of Texas. Anyways we'll have to see
how they cross-connect those Opterons too.

Yousuf Khan

"Lee Waun" wrote in message
...
Are opterons the future of desktops or are they just a stop gap.

They are neither. Opterons are server chip, and possibly a workstation chip.

For the desktop market, AMD will introduce the Athlon 64, which is also an
AMD64 architecture chip, but with less cache.

Yousuf Khan

"Gomer Pyle" wrote in message
thlink.net...
On Mon, 23 Jun 2003 13:34:33 -0400, Yousuf Khan wrote:


I am starting to see a definitive shift towards Linux solutions in
places where there was previously a Solaris-only shop. It's just an
observation I've noticed in the last couple of months.

Yousuf Khan


All of us in the Western World also see the solar eclipse ;-)

Is that some kind of a pun, i.e. "solar eclipse"?

Seriously, a few engineering companies I worked in have
surrounded Sun servers with mostly Linux workstations where
in the past they would have bought all Sun workstations.

What I'm saying is that I'm starting to find that people are wanting Unix
admins to also have a Linux background these days. It's no longer enough to
just know Solaris.

Yousuf Khan

Yousuf Khan wrote:

Perhaps they should call them dead-end servers. Too little, too late.

These AMD Opterons can be scaled upto 8-way/system board with no special
glue logic; they simply work that way out of the box because of the
Hypertransport links. Packaging and heat considerations may prevent anyone
from implementing a practical 8-way system board, but that's beside the
point.

So? I'm not impressed. Sun had systems out with more than 8 CPUs
nearly a decade ago. Inteland is still playing catchup.

With whatever existing crossbar connection technology a manufacturer
might already have, they can easily scale beyond the 8-way.

Why don't you get back to us when they have more than 100.
Of course, by that time, Sun will have raised the bar yet again.

-am © 2003

Anthony Mandic wrote:

Yousuf Khan wrote:

Perhaps they should call them dead-end servers. Too little, too late.

These AMD Opterons can be scaled upto 8-way/system board with no special
glue logic; they simply work that way out of the box because of the
Hypertransport links. Packaging and heat considerations may prevent
anyone from implementing a practical 8-way system board, but that's
beside the point.

So? I'm not impressed. Sun had systems out with more than 8 CPUs
nearly a decade ago. Inteland is still playing catchup.

With whatever existing crossbar connection technology a manufacturer
might already have, they can easily scale beyond the 8-way.

Why don't you get back to us when they have more than 100.
Of course, by that time, Sun will have raised the bar yet again.


Is 9632 processors big enough?

ASCI Red is based on Pentium Pro processors and has been out for a few
years.

Come back when Sun have made more than 10,000, but of course by then
somebody who isn't Sun will have raised the bar even more.

Even the SGI Altix system goes to 64 Itanium chips and is a mass market box.
The Sequent Symmetry boxes did 64CPUs a long time before Sun had it.

Sun are actually playing catchup here, not intel land.

P.

Paul S Brown writes:

Anthony Mandic wrote:
Why don't you get back to us when they have more than 100.
Of course, by that time, Sun will have raised the bar yet again.


Is 9632 processors big enough?

ASCI Red is based on Pentium Pro processors and has been out for a few
years.


Where can I buy these? Experimental platform don't count.

Bye, Dragan

--
Dragan Cvetkovic,

To be or not to be is true. G. Boole No it isn't. L. E. J. Brouwer

Paul S Brown wrote:

I know this is a Sparc/Solaris newsgroup, but sometimes the antipathy
towards x86 boxes gets a bit puerile.

Puerile huh? That's too big a word for the Wintel/Lintel kiddies
to assimilate. However, it doesn't stop them acting like trolls,
retards, arsewipes and other assorted dickheads by crossposting to
inappropriate newsgroups trying to start a flame war.

Historically x86 and SPARC have played in different market sectors, and this
shows in the typical design of the hardware.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
You've got that right.

Assuming that Sun has some overriding technical superiority and will conquer
all is a fallacy. Their hardware has gone from being completely bulletproof
(SS20, 3/480 etc) to more than a little problematic (Ultra 5/10/30/60 - all
with build quality issues).

Its funny you should mention the Ultra 5 and 10 et al. Its also
funny that you don't mention why they are problematic. So I'll
mention why. Its because they has puissant PC peripherals in them.
Whack in IDE drives and controllers, PC-type memory, PCI bus
crapola, etc., etc., etc. and watch the quality go down the tube.
Welcome to the wonderful world of quality PC hardware.

The only major thing Sun have going for them is Solaris.

I think you mean "one of the major ...". Of course, not having
that level of quality didn't stop Microsoft and Intel getting
to where they are today.

Solaris is a damned good OS, but Suns current plans look to try and
undermine that. I have been told by a couple of senior level contacts that
they are going to try to accelerate the feature improvement of Solaris.
What this says to me is more frequent Solaris releases.

No, what it means if more features releases with MUs and
hardware revision releases.

Given that they typically only support the last 3 versions (7,8,9 at present)

You might want to check that.

this is going to end up forcing the hands of some of their corporate customers.

I fail to see how.

I work in an environment where it typically takes us 18 months to certify
a major OS release as releaseworthy and build a standardised hardened build.

When you're dealing with Microsoft I don't blame you.

If Sun bring their product release schedule sufficiently far ahead that we
are certifying obsolete software then we are going to inevatably have to look
to another vendor with more software stability for a long term view,

Well, don't look to Microsoft.

as we will *not* upgrade every machine in the estate (500+ in my bit of it,
6000+ in the whole company) every 18 months just because a vendor wants to
add new shiny bits to the OS.

What a strange regime you operate in.

If my company were to stop buying Solaris running systems then I suspect
this would hurt Sun a lot. I believe that a lot of major corporates are in
our situation.

I doubt it. A lot would welcome new features.

The point of this rant? Sun are a good software company. Their hardware is
only so-so these days, and has been getting more so for years.

As I revealed, this is only with the PC quality stuff. The real
stuff is still excellent.

If they decide to force the hands of their customers then they can be as
good as they want - they will not sell product.

That's merely conjecture on your part. If we take Microsoft's
crud as the perfect example, it hasn't hurt them one bit.

-am © 2003