Intel allows politics to get in way of engineering

When Intel allows politics to get in the way of good engineering they will continue to have the problems they are currently having and it will not get any better until they stop playing politics no matter how many good people they have.

The only real solution for leakage at 90 nm currently is SOI but Intel knows that to implement SOI they have to give IBM access to all of their current chipset and bus patents in exchange for theirs. Rather than risk having some potential real competition to its chipsets and processors Intel has tried to make do with inferior technology and that obviously didn't work.

Intel's next move is to metal gates for Tejas (or whatever they will call the process chemistry fixed Prescott) but IBM is every bit as knowledgeable about that as Intel is and not only is IBM now capable of producing strained silicon as Intel does but theirs is a verifiably superior and, to add insult to injury, a patented process. Clearly, as far as semiconductor technology, Intel is losing this race with no hope in site (they have NO WAY to produce 65 nm chips whereas the IBM chemistry works just fine at that line width). So rather than take the fair and equitable exchange IBM offers, all of our technology for all of yours, Intel has decided to shoot itself in the foot.

What they have learned, and it was probably the boys in Israel beating them over the head with the facts every bit as much as AMD beating them in the marketplace, is that they have to redesign their CPU's to do more real work per cycle (read reduce the number of pipeline stages) even if that means a slower clock speed. To that end I would look on Intel to co-opt every one of Opteron's great ideas (built in memory controller, hypertransport etc.) for their next generation processors starting with Potomac.

That having been said, I find Intel's statement to x-bit labs that they will not enable x86-64 on LGA775 Prescott's to be one of the most arrogant and self serving statements of all time. Their assertion that they are doing it to "Protect" the consumer is so patently false as to be loathsome and flies in the face of Intel's actions with respect to x86-32 where they enabled the technology in the chip two years before there was any software to use it.

No, the only entities that this move protects is Intel's Itanium franchise and Microsoft because there won't be massive number of processors out their that could run a desktop version of Linux on iAMD64. It will also cause a huge problem with their OEM's because if they do this then you can no longer rely on the fact that if the chip is LGA775 then it is 64 bit enabled leading to either an eventual massive write-down of inventory and confusion when they start producing the same chips with 64 bits enabled or huge problems of upgrading everyone's BIOS to enable the technology which was disabled in software. Under all circumstances the consumer is getting well and truly screwed here and makes an upgrade to LGA775 and Prescott at 3.6 GHz one of the single stupidest ideas of 2004.

Name, email address supplied

Intel should be afraid of itself, not the INQ

On the one hand you have the Inquirer running a few truthful statements about what a mess Intel has made of its business. On the other hand you have Intel losing around 12 Billion on the Itanic, at least 5 Billion on that other server chip that can't keep up with an Opteron and 5+Billion on the soon-to-be-discontinued Prescott Pentium 4 or 5 or whatever . Now they are dropping back to an augmented PIII. And they choose to be afraid of McGeek! amazing! They should be afraid of whoever runs their damn company.

Best--Art Horn

Big Hole in Intel's Roadmaps

Something that most commentators have missed is that there is a gaping hole in the current reports of Intel's roadmaps. Let us assume that 2007 sees the new line of both x86 (with added 64-bit) and IA-64. But what are Intel going to do about the very lucrative 1U and blade market, which they currently dominate, before then?

I am beginning to doubt that they will ever manage to get the Prescott to 2-CPU 1U or in a blade at an acceptable speed. And the Banias/Dothan are not ISA compatible - all right, few customers will give a damn, but the difference will be visible. And the current Pentium 4 has a very impressive memory bandwidth, which Banias/Dothan doesn't, but it can't be propped up for another 2-3 years, and the previous trick of just shrinking an old design won't work.

There MAY be a plan to hack the Banias/Dothan into a form that it could be packaged as a Prescott for the 1U and blade market, or there may not. This is shaping up to be an Itanic-scale disaster, except that Intel have real competition and it is an existing market. If you prod the right people, I think that they will twitch, but I doubt they will say anything.

Though, if they were prepared to brazen it out, using the Banias/Dothan and not worrying about the instruction set, hyperthreading or even the memory bandwidth would be feasible. They can still make competitive products, but Intel really doesn't like admitting that it has egg on its face.

Still, if they are REALLY twitchy, it indicates that my suspicions of the problem have some solid basis.

Name, email address supplied

Thanks for outsourcing, Big Guys

Hi Mike,

I have a freelance computer tech support business and I just wanted to publicly thank Dell, HP and all the other computer companies that are outsourcing their call centers overseas. Many of my new clients call me after very frustrating experiences with overseas call centers. Their complaints range from having communications problems to spending hours turning off and re-booting their computers at the direction of the tech support persons to no avail.

These same clients also tell me that they are never going to buy another product from these companies. I sure hope the cost savings they are experiencing are worth the lack of customers they will see in the future.

Sincerely,

Curtis Ritchey
PPA Computer Services