Itanium: A cautionary tale

(continued from previous page)

The initial Itanium prospects were impressive. All the major server and operating systems companies jumped on board.

"Originally, Itanium was envisioned as an architecture to replace the entire spectrum, and that turned out to be overly ambitious."

--Rich Marcello, general manager, HP's Business Critical Server group

Sun created a version of Solaris for Itanium in the 1990s. IBM joined with the Santa Cruz Operation and Sequent to combine their Unix products into an Itanium operating system code-named Monterey. Microsoft offered Windows 2000 for Itanium. Linux allies banded with Intel and server makers in a project called Trillian to adapt the open-source operating system to the chip. Compaq's Tru64 Unix was up and running on an Itanium software. And Silicon Graphics decided to support Itanium and Linux in preference to its own MIPS processor and Irix operating system.

"The momentum was huge," Gwennap said. "There was this incredible anticipation and expectation that this was going to be the next big thing. Intel was on a roll, and with HP backing them, then other companies started jumping on the bandwagon."

Itanium derailed
Then big problems hit. The first Itanium, code-named Merced, was delayed from 1999 to mid-2000. When it arrived even later, in May 2001, even lowly x86 chips beat it in important performance tests.

When Intel and HP launched the Itanium project, "they thought they had just laid the golden egg," Eunice said. However, "when Merced arrived, it was a turd."

Even HP called Merced a mere "development environment."

The delays forced SGI to extend its MIPS chip family by two generations and cancel its first-generation Itanium system. "We had a product we designed based on the Merced chip which we elected not to take into the market," said Dave Parry, general manager for SGI's server group.

And Sun--admittedly a lukewarm ally that never planned to sell its own Itanium servers--dropped Solaris support in 2000.

Intel got the Itanium train back on the tracks after Merced, doubling performance with "McKinley" in 2002. In 2003, it launched "Madison" with 6MB of on-board cache memory; the next year, it unveiled "Madison 9M" with 9MB of cache and a plan for the 2005 release of the dual-core "Montecito."

"Montecito is a fundamentally new, true dual-core design. It does get significant performance advantages over the previous single-core parts," Glaskowsky said.

Behind the scenes, there had been another Itanium shift. An ambitious future-generation product code-named Tanglewood had been planned with as many as 16 processing cores, according to a source familiar with the plan and a document about the chip seen by CNET News.com. But in December 2003, Intel announced the model would be called Tukwila instead--quietly moving to a more conventional design that had four or more cores, slated for release in 2007.

Retreat to the high end
As Intel grappled to produce desirable Itanium products, it gradually reduced its ambitions until the chip's niche was just high-end systems. Itanium is tailored for "the biggest iron," Pat Gelsinger, senior vice president of the Digital Enterprise Group in charge of the servers, said in a March interview.

"I was the one who initiated that, probably two-and-a-half years ago," Marcello said of the high-end shift. "I don't think you can span the entire sever market with one architecture. Originally, Itanium was envisioned as an architecture to replace the entire spectrum, and that turned out to be overly ambitious."

The new direction diminished Itanium's potential influence. "Each time, it was whittled to a smaller and smaller niche, trying to make it more successful," Krewell said.

In 2004, Intel acknowledged that Itanium shipments weren't meeting the company's goals--it had hoped to double its chip sales total in 2004, from 100,000 in 2003.

[fred dunn]

The Itanic is dead...Long live Itanic...

Intel needs to face the reality of their situation and stop this fiasco. Buyers of Intel products shoould realize that the premium that they are paying for the Intel brand is caused by run-away dead-end projects such as this. This so much reminds me of the RAMBUS push by Intel as they swore that no Pentium 4 chipset was ever going to be built for anything other than RAMBUS. But after the first chipset fiasco and much market loss to their own Pentium III they finally caved-in but not after losing money in the process.
Intel wants to dictate technology to the public rather than being driven by the publics wishes.

Their marketing and engineering folks need to get out more because they are on a fast track to losing the public's trust.

Fred Dunn

[System Tyrant]

Itanium...

could have been a contender. I think had Intel pushed really hard to make the Itanium platform the new platform they could have done it, but instead they kept the x86 and pushed it over Itanium. Then comes AMD with x86-64 and basically put more nails into the Itanium coffin.

I think Intel was it's own worst enemy when it comes to Itanium. I think we could all be buying Itanium processors now had Intel went for it. Of course we probably would have hated them for the change, but we would have all gotten use to it.

[Lolo Gecko]

... more pixie dust, less unobtainium ...

the intel/hp dog and pony show is over; it was good while it lasted; but, a (1) trick x86 act none the less. these guys should hang-up their slide rules and call it quits, while they can still exist with a modicum of dignity. clearly, they can still talk the talk; unfortunately, they can't walk the walk. hang it up guys, you're turning our profession into a circus, which makes the rest of us real EE's look like a sorry bunch of over the hill clowns.

Itanium was not a failure...

... at least in the technical sense. It's a hot
power-hog maybe, but at the time it was designed
power efficiency wasn't a design priority.

The problem with Itanium is specifically that
the designers were aiming for a processor with a
revolutionary way of managing instructions that
would permit novel algorithms to perform with
brutal efficiency. Of course, what Intel
marketed the thing as a revolutionary step in
the evolution of the x86 microprocessor line.

The problem being, of course, it was the former
and not the latter. Itaniums are excellent
processors for a very specific subset of
algorithms. The CPU itself doesn't do all the
instruction reordering and such of its
predecessors, instead relying on the compiler to
do these things for you -- and the compiler
often requires the developer to code algorithms
a certain way to take advantage of the CPUs
abilities (much more so than is necessary on
other CPUs). If you ran existing software on it,
you got mediocre results, and the cost of the
CPU was (and really still is) just plain silly.

I had taken some training from Intel on Itanium
development for scientists. It was very
elightening, and I can really appreciate what a
beautiful piece of work Itanium (and Intel's
development tools) were. But at the same time, I
had to balk. I was not likely to recode the
applications I use day in and day out for
Itanium, and I would be beholden to others to do
the same for me. Even though I do develop some
software, I just wasn't interested in having to
go into that level of detail to get the
performance boost I expected. If I were, I'd
have been writing assembly code.

Itanium is flagging not because it's a bad CPU,
only because for the majority it's not the best
CPU or best CPU for the money given what people
need to run today, and because the barrier to
making software that could make it the best is
getting higher the more software people use.

The best case scenario for Itanium would be a
move to a JVM or CLR explicitly tuned for
Itanium and a broad shift to running VM-based
software. Intel probably recognized that but
weren't prepared to affect that shift on their
own.

[Lolo Gecko]

gosh, that's really swell ...

:)

[ramsci]

UltraSparc T1 will nail Itanium's coffin

yeah, check it out man, les than 80 watts power consumption, 5 times Xeon performance, less foot print servers..wow..why would anyone consider Itanium ? If floating point ops are what you need, then there is always Opteron.

Man, Intel should scrap Itanium to save money

[felgercarbnaysay]

Nope

Nope, T1 has no impact on Itanic becuase it does not run Windows. Microsoft killed Itanic because they decided to go the way of history and choose the 64-bit instruction set that makes the most sense. There was absolutely no reason for Intel to come up with a new 64-bit instruction set other than to screw AMD. It bit them in the tail.

forgot

Oh,,, got it ,,Is Itanium a medal something like TITANIUM? iF SO THAT STUFF IS HARD, and very durable, allmost none destructible.. Thank you ..Garey

Booooriiinnggggg

zzzzzzzzzzzzzzzzzzzzzzzz

[felgercarbnaysay]

Alll about trying to screw AMD

When I heard that Intel was going to come up with a whole new 64-bit instruction set for Itanium rather than just extend the old 32-bit set, I knew they were in trouble. I lived through the 16 to 32-bit change of the late 1980s and remember how Motorola, Nat Semi and Intel did the logical thing and just extended their instruction sets. Intel's big mistake was getting fixated on screwing AMD and the other Intel cloners. They forgot that changing instruction sets also screwed their sofware partners and, most importantly, their customers.

I disagree with a previous poster's belief that Sun's Niagara is the nail in Itanium's coffin. The nail was Microsoft when they said no to 64-bit windows using Itanium's ill concieved instruction set.

[System Tyrant]

Repeating myself.

I'm only going to disagree with you on the Intel trying to screw AMD thing. Intel has many ways of screwing AMD they didn't need to come up with the Itanium to do it.

[qmuser]

Good chip, but over-priced

I work with both Intel Itanium 2 based machines, and AMD Opteron based machines. The Itaniums are about 10% faster than the Opterons. However, the Itaniums cost about 8 times as much as the Opterons.

If Intel priced their Itanium 2 chips competetively, their sales volume would surge.

[Pieter_Infomet]

Simply not true.
We work with massive Data Warehouses?
The Itanium really shunts data fast!
What applications are you comparing ??

DEC Alpha all over again

I read this story and couldn't help but replace Itanium with Alpha. It read the same way... lack of apps, re-compiles, 64 bit but no 32 bit compatability, etc. Intel/HP could have looked at the Alpha saga and learned a lesson.

DEC, then Compaq, continued the company line of "Alpha forever", just as Intel/HP are continuing the "Itanium forever" company line. But Itanium will suffer the same fate.

And folks like me will buy AMD Opteron and love the performance, ease of use, and happy users.

[madhungarian]

At least Apha wa real...

Wrong...Alpha was a product! Done delivered and a 64bit processor WAY before it's time. The fact that Qmpaq was not able to make it work had to do with it?s own buy out by guess who... HP! Hell, Intel can't even do a 64 bit processor two decades later? Trust me, it has to do with a cultural impotence there more than anything else and unless Otellini has the balls to change that nothing else will change. That culture could not deliver a baby in nine years!

[kakphoto]

Why don't they just give and admit Defeat

While back in the early days it might have been a good thought, today AMD is the only way to go.
I am old VAX/VMS operator and migrated to the Alphas using OPEN VMS and TRU64. Even the older 400 or so MHZ processors can out do even the newer Servers today. They can run forever without rebooting. Most Windows servers need to be rebooted Daily or weekly even Server 2003.
Then the IBM AS/400's need to be Booted (IPL'd) often.
Yes The Alpha's were 64bit, But they were not meant for Workstation use. They were meant to crunch numbers, and the did that very well.
When Compaq bought DEC, Intel got the rights for the ALPHA, but they still couldn't get the ITANIUM to work. AMD actually hired the People who created the ALPHA chip. Its no wonder why they are overtaking INTEL.

[dcs0582]

Author needs dark glasses and a white cane

Considering that Itanium (HP) is now kicking the collective rears of Sun and IBM, I wouldn't put the author in the same class as Nostradamus. The absolute highest Oracle performance known to man can only be had on IA64 servers. Add overall flexibility and management to the mix, and it's easy to understand why Sun and IBM are terrified of HP's current server lineup.

[Pieter_Infomet]

Most of these comments indicate exactly how unscientific most IT people really are.
The comments here reflect opinions, perceptions, markerting hype and blatant dis information.
They represent very little facts. I find it strange that so many offer opinions when in fact they have not even worked with, or properly benchmarked these machines/technologies they are criticizing (more like character assasination).

This is really not useful. I am not reading these articiles because I am interested in personal opinions or infights.

I hope to gain knowledge and insights from other people's experience.

In stead I am appalled by the biased and unscientific non-sensical opinions and stereo-types
of people who call themselves professionals.

This is simply scandalous!