SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : Intel Corporation (INTC) -- Ignore unavailable to you. Want to Upgrade?


To: Elmer who wrote (140834)8/4/2001 12:50:16 AM
From: Paul Engel  Read Replies (1) | Respond to of 186894
 
Elmer - An EXCELLENT article on Compaq's decision to switch all their high end servers to Intel's ITanium - ESPECIALLY their Himalaya series.

Compaq wins support for processor shift

By Ken Popovich, eWEEK
August 1, 2001 9:36 AM PT

URL: zdnet.com

When Compaq Computer Corp. announced in June that it was redrawing the road map for its NonStop Himalaya servers, executives who oversee the NASDAQ's trading system took notice. After all, they relied on those servers to handle more than $20 trillion worth of stock trades last year.

In fact, Himalaya servers, which can cost millions of dollars each, scale up to 4,080 processors and are valued for their reliability, power some of the world's largest businesses, including 14 of the 15 largest stock exchanges, 80 percent of ATM transactions and nearly 70 percent of all credit card transactions.

So when Compaq announced it would not be migrating the current MIPS-based Himalayas to its veteran Alpha processors as planned, but instead would adopt Intel Corp.'s just-released Itanium, the computer maker risked alienating some of its most prized customers.

But judging by NASDAQ's response and that of other Himalaya users, Compaq has thus far succeeded in winning early support for its decision.

"From my standpoint, it's a net positive," said Steven Randich, chief technology officer for NASDAQ, which uses 50 Himalaya servers to operate its core stock transaction business that has the capacity to handle more than 6 billion transactions per day.

Randich said his top concerns were that the Himalaya line continue, that the NonStop Kernal operating system it utilizes survive and that future systems be able to handle current applications. Compaq assured him on all three points.

As for the decision to adopt Intel's as-yet-unproven 64-bit chip architecture, which was released in April, he said that wasn't a problem.

"I'm comfortable with Intel. All of our Web sites, our internal surveillance systems, plus all of our networks run on Wintel," Randich said.

Compaq acquired the Himalaya line from Tandem Computer Corp. in 1997. Starting in 2003, Compaq had planned to migrate the servers away from using 64-bit processors by MIPS Technologies Inc. to instead utilize its 64-bit Alpha processors, which it acquired in 1998 from Digital Equipment Corp.

The move to Alpha appeared to draw little concern since the chip, first introduced in 1992, had gained a reputation as a reliable, high-performance processor.

But Intel, while the dominant chip maker for PCs and low-end servers, is a relative newcomer to the risk-averse high-end computing market.

Facing the inevitable

At least one Himalaya user admitted that, initially, he wasn't enthusiastic about using Intel rather than the Alpha.

Craig Foote, vice president of ASP Strategy for ALLTEL Financial Services, said he associated Intel processors with problems related to unreliable personal computers and low-end servers.

"But then I realized that it's not Intel processors that people have problems with, it's the operating systems or application programs that tend to fail on Intel boxes, not the processors themselves," said Foote.

The company, which serves more the 24 million telephone customers in the United States, has relied on Himalaya servers for more than seven years, and Foote expects that will continue through the migration.

Migrating to new technology is inevitable, and beneficial, said James Kasdorf, director of special projects at the Pittsburgh Supercomputing Center.

"When Cray Research introduced their highly successful T3D/T3E line, no one questioned their choice of Alpha processors, because Cray was adding significant value at the system level and Alphas were clearly the best available processing element for the job," Kasdorf said. "Compaq is now making a very similar decision, and we expect a very similar result."

Even if companies were wary of Compaq's decision, one industry analyst said, they would be hard-pressed to abandon an architecture in which they've likely invested millions of dollars and that runs their core business applications.

"It would be extremely difficult and incredibly costly," said Rob Enderle of the Giga Information Group. "People would jump out the window before letting them yank those systems out."

And while Compaq's decision to go with Itanium may have surprised some coming just weeks after the chip's launch, Enderle said the computer maker had long planned to phase out Alpha and exit the chip making business it acquired in 1998.

"Itanium was initially supposed to be out in '99, so the extra delay until now did cause a certain pain and force them to retrench and support Alpha for the short term," Enderle said. "But now they're back on track."



To: Elmer who wrote (140834)8/4/2001 1:12:35 AM
From: THE WATSONYOUTH  Read Replies (1) | Respond to of 186894
 
I've seen substantial evidence to support your claim. AMD will see a die shrink going to .13u but not much else.


Just exactly what have you seen??

THE WATSONYOUTH



To: Elmer who wrote (140834)8/4/2001 11:35:41 AM
From: Dan3  Read Replies (1) | Respond to of 186894
 
Re: AMD will see a die shrink going to .13u but not much else

AMD introduced their standard copper process about two years ago (a process equivalent to the one Intel introduced this month, but using the lithography equipment of the period). And AMD's market share has gone from around 12% to around 24% in those two years.

Having spent 2 years working on a new SOI process technology, while Intel was working on catching up to AMD's old plain copper process technology, SOI will let AMD once again jump ahead of Intel, and that will take AMD's market share from 24% to 36% over the next two years.

While just about doubling market share, AMD also increased its relative ASPs from about 1/3 of Intel's to about 2/3 of Intel's. Moving to SOI should also bring AMD ASPs into parity or better with Intel.

The beauty of it all, is that AMD has done this without greatly increasing its costs, so that it is able to do pretty well even in a horrible semi market that coincides with that brief 6 month period when the FAB technology of AMD and Intel overlap. After next quarter, when AMD pulls ahead again, Intel will be left with high costs and low revenues, while AMD will continue to have low costs, but will also have high revenues.



To: Elmer who wrote (140834)8/4/2001 12:13:52 PM
From: muzosi  Read Replies (2) | Respond to of 186894
 
just have a question: why is it that with 250 nm steppers AMD can have a 180 nm process and 130 gates and not have 100 nm gates (or less) with a 198 nm stepper ? IOW why don't you expect AMD gates to scale less than 130 nm when the process goes to 130 nm ? Why should only some layers scale down ?