Interesting stuff! I'm used to stories like this followed by a technology breakthrough allowing Moores Law to continue for a few more years but perhaps we reall are approaching a significant point in chip production. It sounds like one of the main problems is the transistor leakage in current designs preventing mass production of smaller than 28nm wafers, hopefully they can solve these production issues.

We really need to be looking at other materials. They do exist.

We really need to be looking at other materials. They do exist.

True, but in the meantime, companies like AMD/TSMC can finally catch up. On the other hand, intel obviously has the money to make a factory to go the next size smaller, the problem is will they make a product that makes such a die shrink worth doing? If you already own an i7 or an overclocked i5, it's getting increasingly harder to justify an upgrade.

True, but in the meantime, companies like AMD/TSMC can finally catch up. ...

Only if Intel stops. :banana:

14nm broadwell cpu's will be here end of the year

Hang on, AFAIK the law states that transistor count/cpu power will double every, what 18 months or so? Not that die size will shrink.

I'm a little confused. Unless I'm missing something, for AMD/Intel CPU's there is no 28nm CPU's. The current standard is 22nm. Small then 28nm, and I see no indication of that size being non-cost effective.

Is Intel's 22 nm ACTUALLY 22 nm?

Current GPUs are 28nm (AMD 200 series, NVIDIA 700 series). Next is 20nm (Maxwell and AMD Pirate Islands). Current Intel CPUs are 22 nm, Broadwell will be 14nm. http://en.wikipedia.org/wiki/Intel_Tick-Tock and see this: http://en.wikipedia.org/wiki/Die_shrink

I know GPU's are. But when we think of the tick-tock die shrink and moores law, we think of CPU's. The 22nm CPU's have been doing just fine. I'm just trying to understand the article better. Unless it's like BLEH says and it's all a lie. IDK.

Intel made 22nm cost effective by moving to 3D Tri-Gate/FinFET transistor technology. AMD is moving to that with both Global Foundries & TSMC at 16/14nm. They need to move to 450mm wafers to drive costs down. Nvidia has been asking for this for a while (http://www.eetimes.com/document.asp?doc_id=1261542). It's becoming more and more costly to shrink these things. Edit: You guys also have to understand that Intel is like god mode of semiconductor manufacturing. They don't even start using new nodes in mass production till they have 90%+ yields. TSMC on the other hand chargers customers by the wafer, regardless of the yield. Honestly I'm surprised AMD has been able to keep it's CPU's so price competitive for so long, Intel's manufacturing is like light years ahead. That TSMC pricing structure is also the reason why NVIDA prices their chips so high (like the new Titan Z). GK110 is huge and with that comes a higher chance to get a bad chip. Nvidia is paying the same price regardless of what comes out at the end.

I think the law could stand a lot, lot longer. The question is will we be able to afford it. They could do it but i doubt whether the consumer will ever see it in home PC's.