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 (155873)1/16/2002 10:00:35 PM
From: Tony Viola  Read Replies (2) | Respond to of 186894
 
Elmer, all, how about a change of pace from arguing with the chest thumper from droid-land:

HP: McKinley servers will beat Merced

They're talking about 2U servers, but don't say how many processors.

HP is designing a chipset to work with either McKinley or their 64-bit PA-RISC processors. Article talks about much better price/performance vs. Merced, twice the performance at half the price.

HP: McKinley servers will beat Merced
By Matt Loney
Staff Writer, CNET News.com
January 16, 2002, 2:25 p.m. PT
DUBLIN, Ireland--Servers based on the second iteration of Intel's 64-bit Itanium processor, code-named McKinley, will offer twice the performance at half the price of first-generation machines, according to an executive at Hewlett-Packard, which is developing a new McKinley-friendly chipset.

At a press event here this week, HP's Itanium program manager, Marc Botherel, said a decrease in chip prices together with advances in other hardware will have a massive effect on pricing. McKinley is due to ship toward the end of the first half of 2002, although demonstration units are already being installed inside corporations.

Chief among the hardware developments will be HP's "Pluto" chipset, which will support the company's own 64-bit PA-RISC processors as well as McKinley.

"Pluto will be cheap to produce," said Bothorel. "The processors will be cheaper too, but the big savings will come from sharing designs between commercial servers and technical workstations."

Workstations and servers tend to use different chipsets and so need different motherboards, a situation that can raise prices. Pluto, though, will have an AGP graphics bus, a direct connection between the graphics chip and the main processor, said Bothorel. This means that it can be used in workstations, which come with complex graphics chips, as well as in servers, which don't really need much in the way of graphics.

"Pluto will be the first 64-bit chipset to support AGP," Bothorel said.

Ambidexterity in chipsets is a growing trend. IBM's Summit chipset works with the 64-bit McKinley and with standard Xeon processors, which can only handle data in 32-bit chunks.

Such internal architectural changes are likely to help Itanium find a wider audience. The first Itanium processor, based on the Merced core, received a lukewarm welcome from customers. Only 2,600 Itanium servers shipped in the second and third quarter, according to Gartner, while IDC puts the figure at closer to 500. Most of the companies that have bought Itanium systems have been using them in pilot projects.

Thomas Ullrich, marketing manager for Unix systems at HP, said McKinley will fare better because it will have a full software stack. "People will start to push it out in production environments," said Ullrich. "It will take a year to really build up credibility. In 2003 we will start to see broad market adoption."

Costs will also drop as system sizes decrease. Manufacturers will be able to produce McKinley servers that are significantly smaller than servers based on the earlier version of Itanium, the "Merced." HP rack-mounted servers based on Itanium are 7U high, but when McKinley servers go into production, HP will be selling them in 2U packages. One U equals 1.75 inches.

Some of the speed increases meanwhile will come from the McKinley processor itself. This processor will have a 128-bit memory bus instead of the 64-bit bus of the Merced Itanium. The cache will also move onto the die for McKinley.

Matt Loney reported from Dublin.

news.cnet.com



To: Elmer who wrote (155873)1/16/2002 10:10:30 PM
From: wanna_bmw  Read Replies (2) | Respond to of 186894
 
Elmer, Re: "How is AMD going to suddenly bring up .13u when .18u yields so badly?"

Whether AMD has good or bad yields, it doesn't matter. They will still be able to produce more CPUs than they do now, and if Intel doesn't cleverly market the Pentium 4, then they really will lose ASPs. One positive thing is that we know that Intel can hurt AMD to the point where they give up market share for revenues. But AMD's .13u process will be a big cost savings for them, and allow them to break even at lower prices. I also felt strongly positive about their comments on server infrastructure. They are getting the design wins, and it's only a matter of time before a major tier-1 OEM picks up on that. Intel is simply going to have to deliver products that beat AMD's, and they'll also need the cost structure to make a convincing argument about their architecture. Northwood is the first step, and from the tone at the AMD conference, they seem ready to give up on the desktop market, and give Intel a lot of market share back. Where they will profit is from gains in the higher margin server and mobile areas, where Intel still seems to be stumbling. And when it gets to be Hammer Time, Intel better have something, or else.

wbmw



To: Elmer who wrote (155873)1/17/2002 12:26:50 AM
From: Jim McMannis  Read Replies (1) | Respond to of 186894
 
RE:"How is AMD going to suddenly bring up .13u when .18u yields so badly?"

You may be right but yields improved when AMD ditched .35 and jumped to .25 with the K6-2.

Jim