//flex table opened by JP

Click to See Complete Forum and Search --> : Has Intel Broken the Heat Barrier?


Baddog
11-06-2003, 03:01 PM
Has Intel Broken the Heat Barrier?
Putting pressure on rivals, the tech giant has announced a new technology for producing far faster -- and cooler -- chips


In recent years, semiconductor companies have hit a wall when it comes to increasing chip speeds and keeping the heat down. For decades, chipmakers managed to cram more circuits onto a chip, doubling the performance every 18 months. But as the circuitry shrank, it became harder to prevent massive amounts of electricity from overheating computer systems.

Now, Intel (INTC ) says it has an answer. On Nov. 5, the chipmaker reported at a Tokyo conference that it has found a new type of electricity-conducting material that radically reduces the excess-heat problem. Intel now predicts it will be able to continue boosting chip performance for at least 10 more years.

What's the new material? Intel isn't saying for competitive reasons, but it believes it can start turning out the chips in volume by 2007.

MANY APPROACHES. The news is a dramatic twist in the race among top chipmakers to solve one of the industry's biggest problems. Rivals such as Motorola (MOT ), Texas Instruments (TXN ), IBM (IBM ), and others have realized that the current insulating material, silicon dioxide, is too thick to work well as circuits get smaller. Many believe the answer is in switching to a type of metal, although each outfit is following its own R&D path.

Intel also has been experimenting with changing the layout of its chip to shuttle electricity through the circuitry more effectively. So-called trigate transistor chips, which have two sides and a top, are another technology high on Intel's research list.

With the Santa Clara (Calif.) chipmaker planning to spend about $4 billion annually over the next few years to advance research into chip dynamics, the results could have massive implications on everything from PCs to consumer electronics. Chips manufactured with any of these new technologies promise to be a 100 times faster than those on the market today, opening up the door for mass production of tiny devices featuring voice recognition, artificial intelligence, and other hot new technologies. That's good news for whoever gets there first -- and Intel may be in the lead.
http://netscape.businessweek.com/technology/content/nov2003/tc2003116_4281_tc024.htm :t

DaveLeclerc
11-06-2003, 03:13 PM
hmmmmm.
I read in a mag once that when Intel was first trying to move to .9micron they were having a problem with "quantom tunneling". Which is that the electrons moving through the gates on the CPU tend to disappear and reapperar in a diferent gate.
It was in Maximum PC about two months ago.

Someone Stupid
11-06-2003, 04:54 PM
It it isn't out until 2007, they haven't solved it. It's like everything else you hear that will be out in several years from now. They have what they believe is a possible solution that works in a controlled setting most likely scaled up a bit since they are "pushing it back" so far, but it hasn't been proven to the point that I'd say it is a sure thing. It has a good chance of it, but with it being so far back, implementation is a problem, and if you can't implement something, what good is it? Intel will have to deal with heat for the Prescott and Tejas some other way it seems.

bob05
11-06-2003, 06:12 PM
Originally posted by DaveLeclerc
hmmmmm.
I read in a mag once that when Intel was first trying to move to .9micron they were having a problem with "quantom tunneling". Which is that the electrons moving through the gates on the CPU tend to disappear and reapperar in a diferent gate.
It was in Maximum PC about two months ago.

Maybe CPU's now will have to have some form of error correcting on them (either on the CPU itself or on the motherboard).

$1500-P4 gamer
11-06-2003, 07:39 PM
It aint that simple with such INTENSE erata! That chip is pumping data here and magically apearing over there! Thats super bad and totaly unpredictable. How can a error correcting circuit of any type catch and fix that. Best it can do is throw that chunk of data away and request it again (like P4 does) then like what one of every 10 cycles is useless. Oh lol its the P4 willy all over again!;)

Magua
11-06-2003, 11:49 PM
My initial reaction was exactly like SS said. If they have the new material, it wouldn't take 4 years to make a chip.