Intel needs to do that to keep parity with their shitty designs, good let them do it. They'll have all the mistakes, yield problems, and costs that the rest of the industry will learn from. Who Cares?
In CPU industry you have to think to the next coming moves a long time in advance. For example the Crolles fab (SMT, Philips and MOTOROLA) was designed to be able to reach 45 nm, and started at 90 nm.
The Crolles fab was designed when chips where fabbed at 180 nm. Sure that the lessons of 90 nm problems will serve for the next process, but you have to do research on smaller process a long time in advance. That's why only the big guys can afford to stay in the fab competition.
IBM made 20nm chips years ago. The trouble is that the bounds of photolithographic etching are being reached, and in order to get down into the 20nm area, different fabrication methods need to become practical for volume production.
Comments
Now the companies are trying for 65nm?
Shouldn't they perfect the 90nm process first?
Mike
Originally posted by MPMoriarty
Maybe I don't understand a lot about the manufacturing of CPUs, but all I hear is all the problems everybody is having with the 90nm process.
Now the companies are trying for 65nm?
Shouldn't they perfect the 90nm process first?
Mike
there is a thread exactly about this from a few weeks ago, if you look for it.
The Crolles fab was designed when chips where fabbed at 180 nm. Sure that the lessons of 90 nm problems will serve for the next process, but you have to do research on smaller process a long time in advance. That's why only the big guys can afford to stay in the fab competition.
Originally posted by HHogan
http://www.iht.com/articles/536431.htm
IBM sux0rs, why isn't it at 20nm! Damn Macs
IBM made 20nm chips years ago. The trouble is that the bounds of photolithographic etching are being reached, and in order to get down into the 20nm area, different fabrication methods need to become practical for volume production.