HP withdraws Itanium workstations

[David Svensson]

Without a volume desktop market, I think the Itanium will probably go
the way of the 860.

While ofcourse anything can happen. Comparisons with 860, 960 or other
earlier attempts with Itanium are not really valid. 860 went nowhere
and had no impact at all and the difference in investments with
Itanium are huge. Intel realized in 1997-2000 that Itanium would not
be a desktop product. If that was not OK, they would have dropped it
in 2000-2001. Since then they have marketed it as a high-end product
and unless Intel will get in financial trouble I guess they will
continue with Itanium.

 

[George Macdonald]

So it's just a server product now? Boy did Intel screw this one up. The
costs aren't too prohibitive. I wonder why it just didn't take? There is
decent enough reason to need 64-bit at the workstation level. Desktop PCs
just don't matter in terms of 64-bit right now. Workstations you would
think would have taken to it. Intel must have some serious marketing issues
going on.

I don't recall who posted this originally, several months ago, but
"decertification" sounds kinda final:
http://www.ptc.com/partners/hardware/current/itanium_letter.htm

Err, PTC is a pretty important supplier of workstation apps... like
Pro/Engineer. I can make a guess at where this policy comes from: many
software vendors got badly burned by some of the workstation ports to
Risc6K, Alpha, etc. a few years ago. Basically the volume was tiny, income
didn't even come close to covering the cost of buying the hardware and the
majority of customers were happy with x86 based performance... which in
some cases caught up with the workstation's within a matter of months
anyway.

Rgds, George Macdonald

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

 

[The little lost angel]

Did it come that long after IA64? The number of Opterons that have
been shipped so far outstride the number of IA64 that the number of
actual work hours the Opteron had proven itself in has got to be way
higher than IA64.
--
L.Angel: I'm looking for web design work.
If you need basic to med complexity webpages at affordable rates, email me
Standard HTML, SHTML, MySQL + PHP or ASP, Javascript.
If you really want, FrontPage & DreamWeaver too.
But keep in mind you pay extra bandwidth for their bloated code

 

[John Savard]

To the tune of 100,000+... I don't believe that.

It should be noted that an Itanium server, unlike an Itanium
workstation, may contain more than one CPU.

Thus, the fact that these rackmount servers have a high dollar cost is
balanced out.

John Savard
http://home.ecn.ab.ca/~jsavard/index.html

 

[John Savard]

Opteron is proven? Ridiculous comment seeing how it's come long after IA64.
Sparc and Power I agree with but their costs are prohibitive.

Itaniums certainly aren't "unproven" in the sense that there's any doubt
about whether or not the chips *work*.

In fact, I would be rather surprised if both the Itanium and Opteron
architectures haven't been quite well benchmarked.

What, then, could be "unproven" about the Itanium?

I think what the poster means is this: the Opteron uses the architecture
of Intel's Pentium, with some extensions tacked on. So it inherits the
"proven" nature of the architecture of Intel's 80386 chip. Proven to run
Windows 3.1 and Windows 95 and Windows 98 and Windows Millenium Edition
and Windows NT 3.5 and Windows NT 4.0 and Windows 2000 and Windows XP
and simply gazillions of applications for them.

Compared to that, the Power PC is unproven, since the closest thing to a
major platform it is found in is something called a Macintosh... you may
have heard of it.

The trouble, of course, is with such high standards for what constitutes
a "proven" architecture, we would probably all still be using Z-80s.

John Savard
http://home.ecn.ab.ca/~jsavard/index.html

 

[Rob Stow]

Itaniums certainly aren't "unproven" in the sense that there's any doubt
about whether or not the chips *work*.

In fact, I would be rather surprised if both the Itanium and Opteron
architectures haven't been quite well benchmarked.

What, then, could be "unproven" about the Itanium?
I think what the poster means is this: the Opteron uses the architecture
of Intel's Pentium, with some extensions tacked on.

Way off base. The Opteron and Pentium architectures
differ like Antarctica and Hawaii. For that matter,
the P4 and Pentium architectures differ just as much.
Things have changed a *lot* since the Pentium.

So it inherits the
"proven" nature of the architecture of Intel's 80386 chip. Proven to run
Windows 3.1 and Windows 95 and Windows 98 and Windows Millenium Edition
and Windows NT 3.5 and Windows NT 4.0 and Windows 2000 and Windows XP
and simply gazillions of applications for them.

Compared to that, the Power PC is unproven, since the closest thing to a
major platform it is found in is something called a Macintosh... you may
have heard of it.

And just what do you think IBM is doing with all
the processors in the Power line that they don't
sell to Apple ? You don't think big iron from
IBM is a "major platform" ?

 

[Judd]

x86 isn't proven?

Are we talking 64bit?

 

[Judd]

Did it come that long after IA64? The number of Opterons that have
been shipped so far outstride the number of IA64 that the number of
actual work hours the Opteron had proven itself in has got to be way
higher than IA64.
--
L.Angel: I'm looking for web design work.
If you need basic to med complexity webpages at affordable rates, email me
Standard HTML, SHTML, MySQL + PHP or ASP, Javascript.
If you really want, FrontPage & DreamWeaver too.
But keep in mind you pay extra bandwidth for their bloated code

Chrysler sold a lot of K-cars. That doesn't mean they were more reliable
than Mercedes Benz's.

 

[keith]

Are we talking 64bit?

Doesn't really matter much. It's a *small* extension on IA32. ...one
that Intel didn't dare go for to try to (once again) segment the market.
Intel should have stuck with its bread-n-butter. It doesn't do well
outside its corner. AMD has now horned in ot the bread and has a good
chance of taking the butter too.

 

[Yousuf Khan]

Chrysler sold a lot of K-cars. That doesn't mean they were more
reliable than Mercedes Benz's.

I got some news for you about Mercedes Benz's as well, they are about as
reliable as K-cars.

Somebody on the F1 newsgroup put it humourously, "European cars are just
road jewelery". Meaning they look great, cost a lot, and are extremely
delicate.

Yousuf Khan

 

[Yousuf Khan]

Are we talking 64bit?

I don't know, maybe we are, nothing was explicitly stated. One of the
advantages of having a dual 32/64-bit personality is that if you're
considered proven on 32-bit, when people do decide to go 64-bit on you,
you're much further along the proven path there too.

Yousuf Khan

 

[CJT]

Chrysler sold a lot of K-cars. That doesn't mean they were more reliable
than Mercedes Benz's.

Mercedes Benz sold a lot of cars based on the perception they were more
reliable than K-cars. That doesn't necessarily mean they were.

 

[CJT]

I got some news for you about Mercedes Benz's as well, they are about as
reliable as K-cars.

Somebody on the F1 newsgroup put it humourously, "European cars are just
road jewelery". Meaning they look great, cost a lot, and are extremely
delicate.

Yousuf Khan

Oops, I see you beat me to it.

 

[Carlo Razzeto]

If they had announced that they are replacing Itanium workstations with
Opteron workstations, then there would be hell to pay with Intel. Here
they are being diplomatic and stating that they are replacing one Intel
product with another.

Yousuf Khan

Like HP idn't have as much to lose with Itainium? Well perhaps not, they got
what they wanted out of it... Still, it's not like HP doesnt' have an
interest in the chip.

Carlo

 

[chrisv]

Well, the statement is that 100,000+ Itanium processors were "shipped".
There was no mention of them actually being sold. Might be a lot of systems
lent out to developers.

Or shipped to a distributor somewhere, only to be returned on a
stock-rotation some time later... 8)

 

[Yousuf Khan]

Oops, I see you beat me to it.

The more, the merrier.

This one guy I used to work with scrimped and saved and bought himself a
Mercedes SLK convertible. Says it's one of the worst cars he's ever owned.

Yousuf Khan

 

[Mark Hahn]

Without a volume desktop market, I think the Itanium will probably go

While ofcourse anything can happen. Comparisons with 860, 960 or other
earlier attempts with Itanium are not really valid. 860 went nowhere
and had no impact at all and the difference in investments with
Itanium are huge. Intel realized in 1997-2000 that Itanium would not

ah, OK, how about the i432? iirc Intel was pretty firmly committed
to that path as well.

be a desktop product. If that was not OK, they would have dropped it
in 2000-2001. Since then they have marketed it as a high-end product
and unless Intel will get in financial trouble I guess they will
continue with Itanium.

that would be stupid unless they thought they could make it pay;
smart business is ego-less. at this point, there has to be a lot of
questioning within Intel about whether it'll ever all pay off.

for instance, montecito sounds pretty decent, but if it has to wait
for 65 nm in order to be even vaguely manufacturable, well, that's a problem.
personally, I think ia64 will be most profitable to business authors who
will endlessly discuss what went wrong.

 

[keith]

ah, OK, how about the i432? iirc Intel was pretty firmly committed
to that path as well.

Actually, Intel was never "firmly" committed to the iAPX-432. It was born
a dog and died a quick and silent death. ...unlike Itanic that did not
die a quick and silent death. ;-)

that would be stupid unless they thought they could make it pay;
smart business is ego-less. at this point, there has to be a lot of
questioning within Intel about whether it'll ever all pay off.

Sunk money deoesn't have to pay off. It's already sunk. The question is
whether it will pay off in the future, and if there is a better place to
put *future* resources. Water/dam/over.

for instance, montecito sounds pretty decent, but if it has to wait
for 65 nm in order to be even vaguely manufacturable, well, that's a problem.
personally, I think ia64 will be most profitable to business authors who
will endlessly discuss what went wrong.

Hmm, a PHB inspired processor to teach the next generation of PHBs how to
do things. ...Interesting. ;-)