So...Whats supposed to come after 11nm?

[RavenSEAL]

Is there some sort of giant breakthrough expected at that point, is 11nm supposed to be uber-exploitable or is that the point at which technology reaches a certain limit? We looking at 2016-2017?

 

[ElFenix]

that's the point at which you'll need a particle accelerator putting traces into doped diamonds to get beyond

 

[aigomorla]

whats after 11nm silicon?

Umm... i think were going to go on a completely new medium, not silicon, but probably carbon.

Carbon nanotechnology has been showing a very very big leap in technology.
Carbon nanotech is also a better electrical conductor then silicon.

Im fairly sure Intel is already messing around with carbon.

 

[Idontcare]

Pathways exist for scaling down to 5nm. Making it economically viable to actually produce your CPU's at those dimensions (cuz you cheap bastards don't want to pay $10k for your CPU) is the challenge.

 

[wuliheron]

At that scale probably self-assembling nanocircuitry which could possibly still be done on silicon to take advantage of the already existing infrastructure.

http://news.harvard.edu/gazette/story/2011/02/what-ultra-tiny-nanocircuits-can-do/

Whatever comes next you can be pretty sure that the existing silicon infrastructure will be around for some time to come unless someone finds a cheap way to convert it to something better.

 

[Mopetar]

We're at 32 nm right now, so by the time we make it down to 11 nm there's going to be a massive increase (approximately eightfold) in the number of transistors we can fit on a chip. A lot of people are already claiming we have more processing power than most people can reasonable use, so now imagine a CPU with 32-64 cores, because that's what could be built on an 11 nm process. Alternatively, they could just stick with 4-8 core processors that are incredibly tiny and sip power compared to what we have today.

If they make it down to 5 nm on silicon that's another fivefold increase over 11 nm, never mind some of the obscene things that are possible with other materials. IBM has shown off 100 GHz graphene transistors. At that speed you light will only travel about 3 mm per cycle.

It might look difficult from today's perspective, but it's merely an engineering problem. Chip manufacturers and humans in general have been solving those for quite a while now. The massive amounts of computational power we'll have before we hit a wall will just make it much easier to climb when we get to it.

 

[drizek]

We'll start measuring TDP in KW.

 

[Castiel]

When is 11nm scheduled? 2016?

2012 - Ivy Bridge
2013 - Haswell
2014 - Rockwell 16nm

What comes in 2015?

 

[mnewsham]

When is 11nm scheduled? 2016?

2012 - Ivy Bridge
2013 - Haswell
2014 - Rockwell 16nm

What comes in 2015?

Intel is sticking on the same node for 2 years, westmere 32nm, Sandy bridge 32nm, Ivy bridge 22nm, Haswell 22nm, Rockwell 16nm, so if all goes according to plans 2015 should be another 16nm.

 

[stargazr]

Graphene ~

well, maybe one or two steps later.........

http://bigthink.com/ideas/24381

 

[Arkadrel]

http://en.wikipedia.org/wiki/Graphene

In February 2010, researchers at IBM reported that they have been able to create graphene transistors with an on and off rate of 100 gigahertz, far exceeding the rates of previous attempts, and exceeding the speed of silicon. The 240 nm graphene transistors made at IBM were made using extant silicon-manufacturing equipment, meaning that for the first time graphene transistors are a conceivable—though still fanciful—replacement for silicon.

~240nm = meh huge ass cpus....
~100 GHz+ speeds.... Hmmmmmm..... yes please.

It sounds like IMB is at the forfront with this new tech,... wonder if we ll ever see IMB desktop cpus.

 

[Idontcare]

http://en.wikipedia.org/wiki/Graphene




~240nm = meh huge ass cpus....
~100 GHz+ speeds.... Hmmmmmm..... yes please.

It sounds like IMB is at the forfront with this new tech,... wonder if we ll ever see IMB desktop cpus.

xtor density is a function of two dimensions of the transistor (that is why it is called planar CMOS).

Because it serves as a great marketing tool, as laypeople we are conditioned to focus on the reported minimum critical dimension of the transistor which is actually called it "length".

The transistor has a second dimension, called its "width" which goes towards determining the total drive current that comes from switching the transistor.

Drive current is one factor that determines a number of attributes of a given circuit, one of those things being the switching speed of the transistor.

The point is that we laypeople have no basis for concluding that a 240nm effective channel length is going to result in a large die as we have no idea the typical width (not length) of a xtor in our modern 3-4GHz CPUs.

On the flipside, getting a xtor to operate at 100GHz is not THE challenge (it is merely A challenge)...THE challenge is getting an entire circuit block, and many of them, operating at elevated clockspeeds.

Last I bothered tracking down the record holder, CMOS had been demonstrated at ~400GHz and that was on 45nm process tech. Graphene has a ways to go to catch up to CMOS in nearly every respect...but it may actually make chips have lower power consumption or cheaper to manufacture (remember its not performance at any cost, its performance at silly low-cost that compels consumers to upgrade...I submit the 990X as point of fact).