Digitimes claims Moores law is dead - 16nm only in 2019?

[saaya]

While Moore's Law continues to march onward, its pace has slowed. Currently, the semiconductor industry is migrating to 32nm, but the migration to 16nm is not expected to arrive until 2019, meaning that it will take 11-years for the semiconductor industry to advance only two process nodes.
http://www.digitimes.com/bits_chips/a20080911VL207.html

very interesting read about the future of ICs and IT!

[BulldogPO]

It will get harder and harder in every step to shrink process.

[Zytek_Fan]

Moore's Law has actually been dead for a while.
It's just been changed constantly, so it can keep relevance, for whatever reason.

[szukalski]

They'll probably find it was exponential after all..

[Cooper]

They'll probably find it was exponential after all..


It was exponential from the first statement, just skewed on the time axis :D

BTW how many times we heard Moore's law is dead ?

[Stukov]

Moore's Law has actually been dead for a while.
It's just been changed constantly, so it can keep relevance, for whatever reason.
Thats exactly what I was going to say.

[xlink]

They'll probably find it was exponential after all..

ehh moore's law WAS based on being exponential or at the very least close to

[Piotrsama]

BTW how many times we heard Moore's law is dead ?
QFT.

[lithpiperpilot]

BTW how many times we heard Moore's law is dead ?

I stopped counting after a thousand.

[szukalski]

Sorry, I was not clear enough, I meant the reward is inversely exponential. As in the gain isn't as large each time.

Moore's law still has a long way to go. But there are limits to how much you can gain each increment along the time scale.

[Dimitriman]

who cares about Moore's Law...

quantum computing hello?? =]

[anubis]

3D chipdesign (http://www.guru3d.com/news/university-shows-14ghz-cube-cpu-prototype/) to the rescue? Could breathe some life into the old "Moores law" horse :) Plus nano-photonics (http://ietele.oxfordjournals.org/cgi/content/short/E91-C/2/131) and other new paths to keep it going.

And lets not forget quantum computing. How much time did it take from Kennedys speech to a man on the moon? ;) 2019, phahh, im gonna have a qPC capable of storing all the information in the world by that time :rofl:

[foch3 -USA-]

Looks like Intel's Tic Toc Cycle is going to die with it. I have a feeling this is going to hurt Intel more than others because they rely so heavily on their process advantage.

[[XC] gomeler]

Looks like Intel's Tic Toc Cycle is going to die with it. I have a feeling this is going to hurt Intel more than others because they rely so heavily on their process advantage.

Even on the same node Intel has proven to have a superior process. Their research in doping and materials research is rather mindblowing. Just think though, AMD will finally be able to catch up to Intel when neither can go any lower :yepp:

[Cooper]

gomeler;3297863']Even on the same node Intel has proven to have a superior process.

Are you sure about that? :D
Remember presshots ?

[Shadowmage]

Are you sure about that? :D
Remember presshots ?

The problem with Prescott is architectural.

There are publications online that even prove this... AMD suffers around a 20-30% clock speed deficiency due to a process disadvantage at the same node. According to the book (which was written by someone at AMD), AMD's K8 could hit close to Prescott clock speeds (similar FO4 delays) but could not do so due to the slower process. You can search for my previous posts to find the exact name of the book: it has something to do with ASIC designs.

[STaRGaZeR]

Are you sure about that? :D
Remember presshots ?

Prescott is not a process ;)

[Spawne32]

thats because we are reaching the end of how far we can push electrically based systems, the next logical step is light, and intel has already been designing light based systems for experimental purposes, its only a matter of time before we see them on the market.

[Cooper]

Prescott is not a process ;)

Immature process causing big leakage is preshots

[STaRGaZeR]

Immature process causing big leakage is preshots

That was not the reason of Prescott's insane power consumption and horrible perf/watt.

[Shintai]

There will be a huge difference I bet in that time on companies. One or 2 will be at the leading edge. Others will be 4-8 years behind.

16nm..I guess we talk 10billion€ or more.

Immature process causing big leakage is preshots

I think you mix a flawed design and a process node. Hence Dothans was superb on the same process.

[Cooper]

Dothans were running at much lower speed / voltage.

[STaRGaZeR]

Dothans were running at much lower speed / voltage.

Don't use lame arguments. Prescott failure was a design failure, not a process failure.

[Shintai]

Dothans were running at much lower speed / voltage.

And Northwoods vs Prescott? Pipeline changes etc? ;)

In short, its not the process.

[AliG]

honestly though, more's law isn't even a real law, its just more of a hypthesis/postulate anyways, so it's not like it matters. It's not going to end the world since it's just a comment, it won't automatically cause intel to start producing crap cpus

[Shadowmage]

Dothans were running at much lower speed / voltage.

And what does that have to do with the manufacturing process?

You've ignored my point that AMD's process is 20%-30% slower than Intel's.

[[XC] riptide]

Moores law was about Cost. The idea of it been anything to do with size shrink is just a logical extrapolation to transister counts/area. Moores law, per se would still be valid if the same size process stayed the same but just halved in price every 18 months. :)

[Cooper]

And what does that have to do with the manufacturing process?

You've ignored my point that AMD's process is 20%-30% slower than Intel's.

Well ASIC isn't a manufacturing process. And as for preshots I should've specified first revision of these chips. going further Intel improved the process and dropped the leakage values quite significantly.

btw you might as well post that link here for the lazy people like me. thank you

[Shadowmage]

Well ASIC isn't a manufacturing process. And as for preshots I should've specified first revision of these chips. going further Intel improved the process and dropped the leakage values quite significantly.

btw you might as well post that link here for the lazy people like me. thank you

I know that ASICs aren't a manufacturing process. It was a book about synthesis vs. customized logic for ASICs.

http://www.xtremesystems.org/forums/showpost.php?p=3171037&postcount=18

EDIT: It seems that they are now charging for the PDF...

[Hornet331]

Well ASIC isn't a manufacturing process. And as for preshots I should've specified first revision of these chips. going further Intel improved the process and dropped the leakage values quite significantly.

btw you might as well post that link here for the lazy people like me. thank you

Pentium M uses the exact same process as P4 -> P1262, cause the next step is P1264 which is 65nm.

Just look up any decent Pentium M review its mentioned there.