Replies to post #18878 on Intel Corp (INTC)

[Dan3]

Re: They've made gains with HyperTransport, Opteron, and x86-64, but none of those will last, and AMD has nothing else in its bag of tricks. I'm making this call because processor and system microarchitectures are maturing. All we've got left are multi-core and integration. Meanwhile, power and heat are now major obstacles that have to be balanced out along w/ manufacturability and the various vectors of performance. And even performance itself is becoming less relevant.

Let's look at your points:

1)They've made gains with HyperTransport

Yes they have, and AMD will continue to make gains due to it's Hypertransport advantage for a couple more years to come since Intel is taking so long to copy it.

2) AMD64

Intel has copied AMD's instruction set, though it took them 2 years to do it - now they can start trying to catch up in the market. There are a number of proven AMD64 platforms out there in their second and third generation. Intel's first attempt has had problems, particularly with dual core.

3) Heat and power

AMD's more advanced process (with SOI and dual strain) produces more efficient parts across the board. Intel does have a well developed mobile platform design that lets a 32-bit Intel chip keep up (in terms of power use, if not performance) with AMD's 64-bit parts, despite Intel's problematic process, but it's a design that will be looking very outdated by the end of the year.

Will Intel improve its parts and process? Of course!

Will Intel's parts 18 months from now be competitive with what AMD's shipping today? I expect they will.

But you claim AMD has nothing new once Intel catches up to where AMD was last year - Note that AMD has had working fin-fet prototypes for 4 years http://www-leti.cea.fr/commun/AR-2004/P/P1-01.pdf http://www.eetimes.com/news/semi/showArticle.jhtml;jsessionid=331BHXIV2J41IQSNDBCSKHSCJUMEKJVN?artic... and will (as they were with copper, SOI, etc.) almost certainly beat Intel to fin-fet by 2 years or more. Fin-fet parts will let AMD deliver faster parts that use less power even than AMD's present dual strained SOI process.

AMD has a big lead on Intel right now, that lets AMD explore some high end technologies that Intel has to ignore while it concentrates on catching up. The big question is, how much further ahead will AMD have moved, a year from now, when Intel catches up to where AMD was 6 months ago?

[Elmer Phud]

Ten

I'm in agreement with you. My view is that AMD's 65nm process is in big trouble and wouldn't be very competitive if it weren't. They have a white elephant fab to pay for and see no way to fill it. Their bag of tricks is running low and debts are coming due.