Intel delays 4GHz chip until next year

Intel won't ship a 4GHz version of the Pentium 4 until the first quarter of 2005, the latest in a series of delays for the chipmaking giant.

Product plans that the Santa Clara, Calif.-based company released to PC makers on Thursday stated that the 4GHz chip, originally slated for the fourth quarter, won't appear until the first quarter of next year. Company spokesman Howard High said the delay will help ensure that the company can deliver an adequate number of chips when the product is launched.

"We are trying to make sure that we improve our execution," High said.

Product glitches and delays have become a persistent problem for Intel in 2004. The problem prompted company CEO Craig Barrett to issue an e-mail to employees warning of the urgent need to tighten manufacturing and operations at Intel.

Prescott and Dothan, the company's current flagship desktop and notebook processors, came to market after delays. Intel also had to recall some Grantsdale chipsets that make it easy to use a PC as an entertainment center.

More recently, a version of the same chipset for notebooks, code-named Alviso, was delayed until the first quarter of 2005. The company also has had sporadic problems in delivering adequate volumes of some of its latest chips at the time of release.

Manufacturing slips are a serious concern for Intel. Several analysts and executives, including Les Vadasz, the company's fourth employee and one of the major figures in its history, have said that Intel has become a giant because of its production prowess. Barrett himself rose through the ranks as a manufacturing and material science expert. A series of manufacturing glitches in 1999 allowed Advanced Micro Devices to gain market share.

The ability to boost chip speeds is also important in the market share war, as it give manufacturers the ability to cut the price of existing chips. Chip speed, measured in megahertz, is not as accurate as a measure of overall performance as it was in earlier years. Manufacturers also emphasize it less in sales pitches. Still, faster speeds do add performance.

Microsoft, the other half of the Wintel alliance, earlier this week said that it had to delay the 64-bit version of Windows XP.

On a brighter note, High also said Intel is beefing up future chips with larger caches and faster buses. Dual-core chips also remain on track for next year.

Intel's "missteps" and the technology press hostility

I am disturbed by the continued, nonchalant dirt being thrown on Intel over last weeks. There is a prevalent lack of understanding on the writers? side. The strongest opinion comes from those who have never designed any high speed circuitry and those who have no a slightest clue about the complexity of the manufacturing process involved. I can tell you one thing for sure: when you deal with 250 ps clock the amount of problems you are facing is staggering. Intel is doing an outstanding job of ploughing through those problems alone. Remember, we are not talking here about a workshop operation. What is at stake is a production process of the most complex products this civilization ever attempted to produce in high volume! It would be only honest if the writers compared Intel's stage of technology with the other companies taking into account the production volume which is at stake. Some writers have a tendency of pointing out to AMD in a manner which would suggest that it is winning over Intel! This is a fantasy! Advice to the writers: try to be fair, look how "successful" the other competitors are. Can you not see that with all those "terrible" missteps Intel is running smoothly around 60% gross margin? I am personally impressed. Think about approximately 150 million microprocessors Intel is shipping per year (approx. 750,000 per working day). How could you possibly imagine that this is going to be 100% without glitches? Personally I would only complain about the recent problem with 900-series chipset which was by the way promptly identified, corrected and the shipments are going to resume soon. The rest is only another indication to me that people who write about technology will become more and more irrelevant if they are not willing to learn.

[Tex Murphy PI]

Intel needs to get its act together

With all due respects, I completely disagree with the writer being too critical of intel.

Four years ago, the market can pretty much count on Intel to deliver on their promises - and for AMD to make the mistakes.

Today, it is just the complete opposite. Intel can be counted to screw things up, and AMD can be counted on to deliver on their roadmaps.

Intel's problems aren't just in execution, it's also in its priorities.

Challenge: More marketing - less engineering.
1) The P4 is the first chip to actually perform worse than its predecessor on a clock cycle basis. This was done to emphasize marketing's desire to sell by GHz, rather than performance.
2) Intel is moving towards multi-processor cores, rather than improving the core itself. Most applications either don't multi-thread, or lend themselves to multi-threading. This makes the MPC chips worthless for anything short of server work.

Solution:
1) Intel has started moving towards the PentiumM architecture that runs a lot slower, but performs a lot more work per clock-cycle. This is a good first step.
2) Intel must continue to improve the processor core, rather than follow the marketting department's mantra of chasing one fad after another. Improving the core will improve the overall performance of the PC since most tasks are not multi-threaded, or cannot be made to work in such a fashion.

Challenge: Hold engineers and managers accountable for gaffes
1) With the exception of the 865/875 chipsets, Intel has continuously stumbled on engineering design problems with their chipsets. Any employee with that bad of a track record would be summarily fired from any company (except for the government). Where's the accountability?

Solution:
1) I beleive in giving people a second chance, but the chipset group really has dropped the ball far too often, and need to be served notice that it's time to start putting some serious work into their products.

Challenge: Intel's focus is way off.
1) Intel employees waste more time on worthless fluff courses such as "Diversity" / "Political Correctness" courses. I can understand the need for cultural awareness in this Global economy - but Intel takes it to a whole new level. Employees are forced to take course after course of "Diversity" classes, month after month. Hey, folks! Any time left for work?

Solution: Pay more attention to work, rather than fluff.
1) Intel should stop wasting their worker's time with endless "Diversity" / "Politically Correct" courses, and get their people to doing their jobs! A lot of people at Intel are culturally aware - no sense in making them drown in it!

Intel has improved a lot from four years ago. That "Intel Arrogance" has disappeared. And they are a lot more forthcoming with their blunders - compared to AMD.

But Intel has to do more to compete more effectively. They can do it, but the question is whether or not management will allow the rank-and-file workers to do their work.