Intel falls further behind as ~0nm transistors created

[paddy-stone]

7 hours ago, bcredeur97 said:

so what do we do after this? I mean... smaller is no longer an option at that point

 

are we just forever stuck?

Well if these make it to production, the sizes would be significantly smaller than they are now, so it wouldn't be a problem for a great many years I'm guessing,IIRC at atomic levels you would be able to fit billions in a space the same size as the head of a pin, so that's not a problem unless they want to get the processor sizes down to a level where you'd need a microscoep to mount the cooler, lol.

[corsairian]

I remember when 90nm was all the buzz. I think the original Xbox 360 was 90nm before they made revisions and got it to 45nm I think. 

[bcredeur97]

23 minutes ago, corsairian said:

I remember when 90nm was all the buzz. I think the original Xbox 360 was 90nm before they made revisions and got it to 45nm I think. 

it's crazy how the jumps back then were like half the size. now it's maybe 10% here and there

 

7nm should be pretty decent tho. half the size of what intel is stuck on lol

[bob345]

Lab results like this don't matter much to the consumer or tech world. Proving a concept with a physics experiment is one thing, actually having something you can mass manufacture is something entirely different.

[PlayStation 2]

8 hours ago, i_build_nanosuits said:

This news has nothing to do with intel so why is intel even in the title?

Clicky clicky, gotta get them views.

[Aldoarip]

2 hours ago, Dan Castellaneta said:

Clicky clicky, gotta get them views.

It's more of a fanboy thing.

[dizmo]

To be fair, Intel's 10nm process is seen as being basically what AMD's 7nm process is. So. They're not doing that badly. 

[Coaxialgamer]

23 hours ago, bcredeur97 said:

so what do we do after this? I mean... smaller is no longer an option at that point

 

are we just forever stuck?

This is when real innovation starts. No longer will shitty designs survive because of process alone. 

In the near future, this is what you can expect :

 

-Manycore

-2.5D and 3D integration 

-further specialization of circuitry, move towards dedicated hardware instead of a general purpose approach. 

-Multi-chip modules 

-Active interposers

 

Also, the focus will be towards improving the efficiency of designs: is it really a good idea to spend half of your transistor and power budget on massive multi-MB caches?  Is it really a good use of ressources to build complex out of order superscalar architectures in pursuit of minor single threaded performance gains? 

[JoostinOnline]

14nm has nothing to do with the size of the transistor. It's the space between transistors. This isn't related to Intel.

[cj09beira]

7 hours ago, dizmo said:

To be fair, Intel's 10nm process is seen as being basically what AMD's 7nm process is. So. They're not doing that badly. 

problem is they will be scaling it back up to improve yields enough to be profitable so they will launch something closer to a 12nm 

[CTR640]

On 16-8-2018 at 5:58 PM, bcredeur97 said:

but it's been covered that those are more for specific purposes rather than general purpose computing like transistors are

Back when printers were developed, only big companies could afford them and no household could buy them Idem dito for computers.

Those 2 techs have made it to households. Same goes for cars. Only the rich could buy cars and now some cars are cheaper than iPhone X, of course it depends which model it is and in which shape but you get the point.

[Coaxialgamer]

16 hours ago, CTR640 said:

Back when printers were developed, only big companies could afford them and no household could buy them Idem dito for computers.

Those 2 techs have made it to households. Same goes for cars. Only the rich could buy cars and now some cars are cheaper than iPhone X, of course it depends which model it is and in which shape but you get the point.

This is different. Printers and computers largely evolved from iterative improvements in manufacturing. It was simply a problem of taking existing tech and cramming it into a smaller space. Sure, there were plenty of innovations along the way, but you catch my drift. Same for cars: reduce cost at the low end, make it better at the high end. Rinse and repeat. 

 

Personal, desktop quantum computing would require nothing short of a breakthrough on a scale we haven't seen since the bell labs transistor or penicillin. This is not something that can be had by just hoping the tech will get better over time. 

[yian88]

Sadly 0nm wont help at all.

What we actually need is a new way of storing RAM, something as fast as L1 cache, currently CPU's are massively slowed by l2/l3 > Ram cache unless the software is perfectly optimized with data driven desing in perfect alignment.

We also need MCM for gpu's and better infinity fabric-like (from AMD) tech for CPU's with lower latency, otherwise monolithic cores will be more and more expensive as fabrication process node gets smallers, the yields will be bad and costs high.

[Pixel5]

this is very very far from being useful for the consumer market and also note that its not as big of a deal for computers as it seems because now you have a very small switch but you still need to connect to it and have lines that can carry the current connected to all of these transistors.

 

so you kind of have the world's smallest traffic light that still needs to manage a giant road that you cant get rid off.

[mr moose]

18 minutes ago, Pixel5 said:

this is very very far from being useful for the consumer market and also note that its not as big of a deal for computers as it seems because now you have a very small switch but you still need to connect to it and have lines that can carry the current connected to all of these transistors.

 

so you kind of have the world's smallest traffic light that still needs to manage a giant road that you cant get rid off.

That is how 5nm started, and 10nm and 14nm and the very first transistor.    Who knows how they will resolve the next issue, but I can assure you if history is anything to go on (constantly proving it wasn't impossible after all) they will resolve it.