Free Republic
Browse · Search
News/Activism
Topics · Post Article

Skip to comments.

Intel To Ship 2nd Itanium Chip To Rising Expectations
Lycos Worldwide ^ | 3 Jul 2002, 5:39pm ET | Mark Boslet, Dow Jones Newswires, 650-496-1366; mark.boslet@dowjones.com

Posted on 07/05/2002 7:06:37 AM PDT by Ernest_at_the_Beach

PALO ALTO, Calif. -(Dow Jones)- If at first you don't succeed, bring out Itanium 2.

That is what Intel Corp. (INTC) will do on Monday to rising expectations that this new top-of-the-line chip will do what its predecessor couldn't: compete for the most demanding of corporate computing jobs. That will mean taking on the titans of high-tech computing, International Business Machines Corp. (IBM), Sun Microsystems Inc. (SUNW) and Hewlett-Packard Co. (HPQ), all of whom make the powerful Unix servers companies rely on for their internal business systems.

Intel is expected to unveiled three Itanium 2 chips, and computer makers such as H-P, IBM, and NEC Corp. (NIPNY) will announce they plan to offer machines using the microprocessors. Noticeably missing from the parade, at least for time being, will be Dell Computer Corp. (DELL), which says it is waiting for Itanium demand to increase before offering servers.

Two of the chips to come out Monday will run at speeds of 1 gigahertz and one at 900 megahertz. Each will sell at prices similar to the company's present Itanium products - which are $1,177 to $4,227 - and boast of about a two-fold performance improvement, analysts say. Intel declined to discuss details of the announcement in advance.

The "very exciting performance" gains "should make this round a lot different from last year," when the first Itanium hit the market to poor reviews, said Nathan Brookwood, a principal at market researcher Insight 64. "We will see a lot more competition among system vendors."

Any improvement in market acceptance would be welcome at Intel as it tries to broaden its business beyond the personal-computer market. The company spent an estimated $1 billion developing the product line and watched as sales of its first major chip redesign in 16 years were unexpectedly sluggish.

"It's a very important product for us," says Lisa Hambrick, director of Intel's enterprise processor marketing. The high-performance server market is dominated by Unix products. "We're looking to change that."

But the hurdles are considerable. Even loyal computer maker Dell is proving slow to line up. "We're taking a wait and see approach to it," says spokesman Bruce Anderson. The company is "really waiting for higher volume." Dell will continue to sell the PowerEdge 7150 it launched last May with the first Itanium chip, but has discontinued an Itanium workstation. It also says it is looking at Intel competitor Advanced Micro Devices Inc.'s (AMD) upcoming Opteron chip, which plans to take on both Intel's Xeon and Itanium server processors.

"We're talking to AMD," says Anderson. "We're evaluating their technology all the time."

AMD plans to ship Opteron in the first half of next year and claims it is an easier step into 64-bit computing than Itanium. Chips with 64-bit technology process data in larger chunks and are more efficient at scooping long strings of information out of big databases and programs.

Opteron is designed to drive down costs and appeal to a broader segment of the server market than Itanium, said AMD Marketing Director Benjamin Williams. However, it also isn't being designed to run in top-end servers that use more than 8 processors at a time. Companies such as H-P hope to sell systems with 64 Itanium 2 chips by next year. Still, according to Williams, Itanium's success is not assured. "I'm not sure it's been adopted."

Some vendors believe that will change. "The marketplace will pick up starting in the second half of the year," says Mark Hudson, worldwide marketing manager at HP. "This is an architecture that will be around for the next 20 years."

Hudson estimates HP's Itanium 2 systems will be about 20% cheaper to buy than the H-P Unix system as well as smaller and less costly to support. "I think people's minds are coming around," he said.

But the "ecosystem" of software developers and service companies needs to expand. "The ecosystem is the biggest question mark," he said.

Intel anticipates growing backing for the chip. The first Itanium had just slightly more than 100 mostly technical commercial programs running on it. A new wave of Itanium 2 programs from big software makers including Oracle Corp. ( ORCL), SAP AG (SAP) and Microsoft Corp. (MSFT) is expected late this year or early next. BEA Systems Inc. (BEAS) will have it application server available in the fourth quarter, says Vice President Gamiel Gran.

The first Itanium's strength was its number-crunching ability, a trait useful in scientific applications. Changes to Itanium 2 now make it better at moving around lots of data, said Brookwood. This is an attribute valued by corporations. Intel will find out if it is enough.

-By Mark Boslet, Dow Jones Newswires, 650-496-1366; mark.boslet@dowjones.com

(This story was originally published by Dow Jones Newswires)


TOPICS: Business/Economy; Extended News; News/Current Events; Technical
KEYWORDS: computing; economy; intel; itanium; techindex
Navigation: use the links below to view more comments.
first 1-2021-26 next last
Maybe this will get Business investment moving again.
1 posted on 07/05/2002 7:06:37 AM PDT by Ernest_at_the_Beach
[ Post Reply | Private Reply | View Replies]

To: *tech_index; Mathlete; Apple Pan Dowdy; grundle; beckett; billorites; ErnBatavia; One More Time; ...
To find all articles tagged or indexed using tech_index

Click here: tech_index

2 posted on 07/05/2002 7:08:04 AM PDT by Ernest_at_the_Beach
[ Post Reply | Private Reply | To 1 | View Replies]

To: Ernest_at_the_Beach
Doubt you will see a major move right away. Most applications will need to be rewritten to take advantage of the 64 bit architecture. Also, with the introduction of clustering technology, most systems today that need high performance, can acheive it by using a less expensive cluster of lower priced chips.

We are in a competing technologies market, clustering vs faster processors. In these types of markets, lower costs, tend to win out. At first, it would appear that clustering would be the lower cost due to the lower cost of initial hardware. However, it often comes at a higher cost in operational overhead. With the 64 bit chips, you have an application re-write overhead. My gut is telling me that managers will adopt a wait and see attitude.

It will be interesting to see how this moves. I suspect that since both technogies are new, it will be a "fair" fight. Just my .02

3 posted on 07/05/2002 7:48:56 AM PDT by taxcontrol
[ Post Reply | Private Reply | To 1 | View Replies]

To: Ernest_at_the_Beach
I'm missing something here. I know some of you computer people are going to bitchslap me, but I thought 1 gigahertz was kind of slow by today's standards? The computer I use has a 333 Celeron (megahertz) My wife bought a new one- Pentium 4 and it runs at some ridiculous speed that I don't even like to contemplate- well over 1 gigahertz (1.8 I think). Obviously, I'm missing something.

Come on, inform me! I have assumed the position and await enlightenment!

4 posted on 07/05/2002 8:08:31 AM PDT by Prodigal Son
[ Post Reply | Private Reply | To 1 | View Replies]

To: Prodigal Son
The Itanium is a 64 bit chip, while PC's have 32 bit chips. A 64 bit chip has a much larger instruction set and address space than a 32 bit chip so, gigahertz aside, it can process more complicated instructions more quickly. A 64 bit chip is more of a minicomputer than a PC, with the capacity to support multiple processes more easily than a 32 bit PC.

This is of particular interest to me because my favorite operating system, OPEN/VMS is being ported to Itanium by HP. VMS is a whole lot more robust than Win 2000 and much easier to use than command line UNIX. Until now, it has been stuck on the dwindling Alpha/VAX minicomputer architecture but this will hopefully give it a new lease on life. If someone could mate OPEN/VMS with a GUI and standard WINTEL applications like Office it would be a really wonderful thing.
5 posted on 07/05/2002 8:42:41 AM PDT by Tokhtamish
[ Post Reply | Private Reply | To 4 | View Replies]

To: Prodigal Son
We have heard over the last few years, first from Apple, then from AMD about the "Megahertz Myth". Basically what this means is that the clock speed - or Mhz" is not an accurate measure of processing power or speed.

The current slate of P3 and P4 processors have truly amazing clock speeds, but their "pipeline" or path the data takes through the processor is restricted because it is so long. Therefore, less data is processed per clock cycle.

AMD processors, and even more so - Motorola G4 processors have shorter pipelines and therefore don't have the same restriction on amount of data/instructions processed per clock cycle. Because of this, more work is done each clock cycle. It's basically an efficiency issue.

Say that processor "A" can execute 8 instructions per clock cycle.

Processor "B" can execute 5 instructions per clock cycle.

Processor "B’s” clock speed is 50% slower than "B". In the same given time frame, theoretically they would actually process the same total quantity of information., yet processor "A" did it more efficiently. Combine this with the fact that "B" is actually going to do a bit less total output because of inefficiencies in the design.

The Itanium line is shows that Intel is finally admitting this fact. While Mhz (or Ghz) sounds great and still sells consumers, professionals what performance, which does not always come directly from clock speed.

Another monkey wrench is that the software really needs to be optimized for the specific processor architecture.
6 posted on 07/05/2002 8:51:59 AM PDT by TheBattman
[ Post Reply | Private Reply | To 4 | View Replies]

To: TheBattman; Tokhtamish
Thanks for those replies. I think I understand now. At least I'm going to tell the wife that and rag her that her computer SUCKS! LOL! Seriously, thanks.
7 posted on 07/05/2002 9:16:58 AM PDT by Prodigal Son
[ Post Reply | Private Reply | To 6 | View Replies]

To: Prodigal Son
Aside from better instruction sets & etc., the pathways of the new chip are twice as wide: 64 instead of 32 bits.

Think of it in reverse:  When Intel was still producing the 32-bit PII, AMD was making the 16-bit K6.  But the performance of the two chips was simular because Intels "300 MHZ PII" was running at 300 MHZ but AMDs "300 MHZ K6" was actually running much faster: more like 550 MHZ.  The net effect was that it ran like a "300 MHZ" PII.

As for day-to-day applications most of the reasons you notice your wife's new computer running so fast is not the chip itself but the peripherals like the hard drive.  You probably have a 3400 RPM drive and you wife's system probably has a 7200 RPM drive with UDMA 100.  She probably has DDR memory too, as opposed to your memory which is probably running at 66 MHZ.  Not to mention a superior video card.

Where the processor really comes into play is in rendering graphics or sorting large database tables...When opening Word or something like that, it's everything else in the computer that comes into play.

8 posted on 07/05/2002 9:29:42 AM PDT by Psycho_Bunny
[ Post Reply | Private Reply | To 4 | View Replies]

To: TheBattman
AMD processors, and even more so - Motorola G4 processors have shorter pipelines and therefore don't have the same restriction on amount of data/instructions processed per clock cycle. Because of this, more work is done each clock cycle. It's basically an efficiency issue.

That's true, as far as it goes. But there's a balance to be struck between clock rate and IPC's. The reason you increase the pipeline is so you can up the clock rate in the first place, so it's a trade-off - it's not the case that a shorter pipeline automatically equals better performance. Really, who cares about "efficiency" over raw performance? It's not like a car, where the mileage means something to the end user ;)

So, the first G4's had a 4-stage pipeline, the G4e's have a 7-stage pipeline, and the G5 is supposed to have a 10-stage pipeline, same as Itanium, so it can be introduced at 1.6 GHz. Which will put Apple/Moto in the difficult position of explaining that the "megahertz myth" applies only to Intel processors ;)

P4 to Itanium is a little different than jumping from G4 to G5, which is probably why it's slow to be accepted - G5 supposedly has a section for 32-bit backwards-compatibility, while Itanium is all 64-bit VLIW processing, which means a recompile for everything, unlike the G5...

9 posted on 07/05/2002 9:35:00 AM PDT by general_re
[ Post Reply | Private Reply | To 6 | View Replies]

To: Ernest_at_the_Beach
bttt
10 posted on 07/05/2002 10:07:36 AM PDT by Free the USA
[ Post Reply | Private Reply | To 2 | View Replies]

To: general_re
Care to guess when the G5s will be released? Surely OSX 10.2 (Jaguar) has been written with the G5 in mind. That should be out later this summer.
11 posted on 07/05/2002 10:22:28 AM PDT by savedbygrace
[ Post Reply | Private Reply | To 9 | View Replies]

To: Psycho_Bunny
When opening Word or something like that, it's everything else in the computer that comes into play.

For a particular task, a computer does work at the speed of the slowest component involved in that work task!

Fast computers and slow computers all wait at the same speed!

The world of the Personal Computers has really come a long way. We are now rehashing many of the (nearly) same arguments that took place relative to mainframes and supercomputers some years ago!

12 posted on 07/05/2002 11:16:46 AM PDT by Ernest_at_the_Beach
[ Post Reply | Private Reply | To 8 | View Replies]

To: savedbygrace
Care to guess when the G5s will be released? Surely OSX 10.2 (Jaguar) has been written with the G5 in mind. That should be out later this summer.

Honestly, I don't know when they'll be out, beyond the obligatory Real Soon Now - sometime this summer or fall, I'd imagine ;)

13 posted on 07/05/2002 11:24:48 AM PDT by general_re
[ Post Reply | Private Reply | To 11 | View Replies]

To: general_re
Intel has been marketing MERCED? ITANIC? ITANIUM?-2?
for almost 10 YEARS!!!It doesn't WORK! Compaq refused
to ship crappy Itanium systems because they failed a stress
test of the CPU--Compaq proved that failure on DELL and IBM
machines! G4--Ultra sparc--Alpha--PA risc--all better archtictures then this (EPIC)from intel. Intel is a
GORRILLA for one reason only their alliance to Microsoft-
If Microsoft gets their Windows to work in 64 bit and lowers
the boom on Oracle with SQL at dirt rate pricing you will see a shift--Now in 1997 Microsoft was demo-ing 64 bit apps.
on high end Alpha machines (8400-12 processor 8 gig memory)
What happened?-Compaq bought DEC-killed NT on Alpha-last
year Killed Alpha altogether sent the engineers to INTEL!!
14 posted on 07/05/2002 1:05:45 PM PDT by mj1234
[ Post Reply | Private Reply | To 9 | View Replies]

To: mj1234
I'll wait for the SPEC scores, myself. FWIW, the Itanium2 is reported to have better unofficial SPECfp scores than the Power4, and that's saying a lot.

Anyway, you're wrong. Alpha isn't dead just yet - HP just released their Alpha roadmap the other day.

15 posted on 07/05/2002 1:26:55 PM PDT by general_re
[ Post Reply | Private Reply | To 14 | View Replies]

To: TheBattman
"AMD processors, and even more so - Motorola G4 processors have shorter pipelines and therefore don't have the same restriction on amount of data/instructions processed per clock cycle. Because of this, more work is done each clock cycle. It's basically an efficiency issue."

I have never trusted AMD chips due to concerns about compatibility. Since the days of DOS and Win 3.1. I used to write and sell graphics software and a user reported a bug I could not reproduce. Fortunately he was located close to me. On a hunch, I looked inside his PC and found a non-Intel chip. Replaced it with a "genuine Intel" and the problem went away. A big piece of my software was in assembly; talked directly to the hardware. I still have such concerns, but perhaps they are now unwarranted. Any advice?

--Boris

16 posted on 07/05/2002 3:37:54 PM PDT by boris
[ Post Reply | Private Reply | To 6 | View Replies]

To: Psycho_Bunny
Uhh, there hasn't been a 16-bit x86 processor since the 286!

The K6-2 and the Pentium II are both 32-bit processors, the K6-2 just used the P54 bus and the Pentium II used the Pentium PRO bus.

The improved performance of the Pentium II over the K6-2 was the result of backside cache and better FPU
17 posted on 07/05/2002 3:54:56 PM PDT by ContentiousObjector
[ Post Reply | Private Reply | To 8 | View Replies]

To: savedbygrace
consitering the fine shape Motorola is in, I would guess the PowerPC 8500 which was supposed to be the "G5" will never see the light of Macworld
18 posted on 07/05/2002 3:56:26 PM PDT by ContentiousObjector
[ Post Reply | Private Reply | To 11 | View Replies]

To: ContentiousObjector
LOL! You're right. I completely spaced that.

But they were run hotter to make up for performance.

19 posted on 07/05/2002 4:00:07 PM PDT by Psycho_Bunny
[ Post Reply | Private Reply | To 17 | View Replies]

To: Ernest_at_the_Beach
Interesting discussions about architecture.

The thing to remember is that neither MHz rating, pipelines, or the number of bits in the word, 32 or 64 is enough by itself. Its how the whole things plays together with everything else. Pipelining for example depends on branch prediction of an conditional statement: if you guess wrong, guess what? you have to THROW AWAY the contents of the pipeline unless you actually have loaded BOTH BRANCHES

Other factors will come into play: how cache is used, how big the cache is and how cache coherence is handled if you have multiple CPUS sharing physical memory...what do you do if one of the computers changes an area of memory and the other one was also accessing it?

It is a complex game and i can hardly wait to see how the new CPUS handle. Like watching Ferrari, Jaguar, Porsche and Mercedes duke it out at LeMans...(if the damn Eurosnots hadn't changed the rules to exclude the GT40 we could probably add Ford to this list)

20 posted on 07/05/2002 4:18:18 PM PDT by chilepepper
[ Post Reply | Private Reply | To 1 | View Replies]


Navigation: use the links below to view more comments.
first 1-2021-26 next last

Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.

Free Republic
Browse · Search
News/Activism
Topics · Post Article

FreeRepublic, LLC, PO BOX 9771, FRESNO, CA 93794
FreeRepublic.com is powered by software copyright 2000-2008 John Robinson