07-08-2003 08:31									#1
	IBM's Push for 65-nm Chip Development Suffering from ads?

http://www.macrumors.com/pages/2003...806185951.shtml

Well, what do you know? I knew that Apple could count on IBM to help them out.

Long live Apple!!!

Long live IBM!!!

Die Motorola, DIE!!!

	07-08-2003 08:33									#2
	Get a bigger avatar today!

2005/2006 eh? Looks like I'll wait a few years to graduate college and go for a G6. from a G4 to a G6, that'll be pretty sweet.

	07-08-2003 08:37									#3
	Avatar Enlargement: We've got the solution

quote: Originally posted by monolith94 2005/2006 eh? Looks like I'll wait a few years to graduate college and go for a G6. from a G4 to a G6, that'll be pretty sweet.

If you read that more carefully you will see that it says that "this is of unknown reliability" (referring to 2005/2006 release date). That could be a good thing or a bad thing.

	07-08-2003 08:38									#4
	Suffering from ads?

By the way, by then I imagine that the G7 or G8 will be out.

	07-08-2003 10:22									#5
	Support Apolyton, buy Galactic Civilizations

IBM's SOI process is a piece of ****: they already screwed over HP and Transmeta, now they are in the process of screwing AMD over. I think Intel is laughing its ass off.

	07-08-2003 10:23									#6
	Support Apolyton, buy Alpha Centauri

DIE APPLE, DIE!

	07-08-2003 10:37									#7
	Help yourself to an AD-FREE life

I see IBM is still playing catchup to Intel.

quote: Originally posted by Lord Merciless IBM's SOI process is a piece of ****: they already screwed over HP and Transmeta, now they are in the process of screwing AMD over. I think Intel is laughing its ass off.

To be fair, Intel avoided SOI and is trying strained silicon instead, and it appears to be just as (in)effective as IBM's SOI. Intel's 90nm process shows almost no improvements in gate-leakage that the strained silicon should've provided.

Intel's keeping one step ahead of everyone in the market though, it seems to be their gameplan -- it's going to mass produce 90nm chips at nearly all its fab plants only a month after IBM/Apple launch their new 130nm chip.

And AMD's not slated to go to 90nm until half-way through 2004...

	07-08-2003 10:48									#8
	Browse Apolyton AD-FREE

These chips are all inferior to ball bearings.

	07-08-2003 18:15									#9
	Support Apolyton, buy Call to Power 2

anything on the path to more advanced technology.

	07-08-2003 19:08									#10
	Support Apolyton, buy Civilization III: Complete

And for the record, this thread is pretty stupid -- all of the chip makers (even Motorola!) want 65nm chips and are heading there. And after that it'll be 45nm.

It's no news whatsoever that IBM says they want to go to 65nm chips...hell, back in January Intel announced it has already produced masks at 65nm and DRAM now as well has been produced.

IBM's kinda behind everyone still, but trying valiantly.

	07-08-2003 19:19									#11
	Support Apolyton or Terrorists Win

And what comes after 45nm???

	07-08-2003 19:22									#12
	Lose 30 kilos (of popups)

35nm probably.

What's worth noting is that the current techniques for achieving those small processes is somewhat unable to do anything below 65nm. Intel's been looking into improvements now for years, and seems to have found it with a technology called Extreme Ultraviolet (EUV) lithography.

There's a cool, non-overly-technical page on this kind of stuff (future of processes) here: http://www.intel.com/labs/features/si02031.htm

Excerpt:

quote: Intel has made major strides in the next generation process known as Extreme Ultraviolet (EUV) lithography. EUV lithography uses a series of mirrors to direct light with a wavelength of 13.4nm to print exceptionally small features below 45nm. Making computer chips using EUV light may lead to microprocessors that are tens of times more powerful than today's most advanced chips, and memory chips with similar increases in storage capacity. Still Working in a World Without Limits Intel recently announced it has successfully fabricated EUV masks for the 32nm-node generation. Since EUV is a new technology, Intel is paralleling its in-house mask-making operations by developing the necessary supply chain with other companies. This cooperation helps ensure the tools, blanks, and other components needed for mask design and manufacture will be available when production starts. EUV lithography is projected to take Intel at least four generations — nearly a decade — beyond its current 0.13-micron process generation. By recognizing masks as a key driver in advancing the future of silicon manufacturing, Intel continues to make the advances and investments necessary to keep them from being a barrier to extending Moore's Law.

(0.13-micron = 130nm, for the non-Scientists )

	07-08-2003 19:29									#13
	Avatar Enlargement: We've got the solution

A major problem now with the continual shrinking of processes is heat dissipation. Each new generation is expontentially worse at "gate leakage", or wasted energy that eventaully just gets dumped off as heat.

It's a minor problem at 130nm, and it's a major one at 90nm. It's why Intel had to delay the upcoming upgrade to the Pentium M (130nm to 90nm), because heat dissipation went up from 24W to 30W. Which is a lot for a mobile processor. Current steppings of Prescott (90nm upgrade to the Pentium 4) are running at 100W, but Intel says it's going to lower that before release in October.

Silicon-on-Insulator (SOI) is what IBM tried to fix this with, and it didn't exactly work.
Strained Silicon is what Intel tried to fix this with, and it hasn't exactly worked (so far), but they haven't written it off yet -- 'cause 90nm is their first process with it and it's brand new still and is being tweaked.

	07-08-2003 23:32									#14
	Full PM-box? Change here!

After 45nm, the "brickwall" approaches, and it's for real this time.

	07-08-2003 23:33									#15
	Support Apolyton, buy Galactic Civilizations

EUV may fix that.

	06-09-2003 01:44									#16
	Browse Apolyton AD-FREE

cool

Jopn Miler

	06-09-2003 02:11									#17
	Full PM-box? Change here!

Bah. I won't settle for anything less than Dexler's supercomputer the size of a human cell.

	06-09-2003 02:13									#18
	Remove this text

diamond chips are the answer.

	06-09-2003 02:17									#19
	Support Apolyton, buy Civilization 2

Diamond's not a semi-conductor iirc. And its tricky to get it to form exactly how you want it.

Carbon nanotubes are semi-conductors if prepared right. Go Carbon!

	06-09-2003 02:45									#20
	Support Apolyton, buy Galactic Civilizations

quote: Originally posted by Asher EUV may fix that.

It's not the lithography, but reduced gain in device performance combined with a slew of horrible short-channel effects that's going to make ultra-tiny device unattractive.