bit-tech.net

Intel details 10nm, 7nm, 5nm process roadmap

Intel details 10nm, 7nm, 5nm process roadmap

As semiconductors move to smaller process nodes, the problems mount up - but Intel is looking to a 2015 unveiling of a 10nm-based chip.

Semiconductor giant Intel has revealed its roadmap for process technologies, which will see 10nm, 7nm and 5nm released beginning in 2015.

In semiconductor manufacturing, process is king: the term refers to how chip designs are shrunk from their giant human-viewable schematics down to teeny-tiny production parts with components several times thinner than a human hair. The better a company's process, the smaller the final chip; the smaller the final chip, the better the performance. Smaller process nodes also mean more chips from a particular silicon wafer, albeit tempered with the usual yield problems as issues with new process nodes are ironed out.

The majority of the industry is working on a 22nm process at present, including Intel's recently-launched Ivy Bridge processors. The next step for the semiconductor industry is 14nm, which Intel is planning to introduce with its Broadwell processors - the successor to the new microarchitecture Haswell parts, based on the same 22nm process as Ivy Bridge as part of Intel's 'tick-tock' development cycle.

All companies in the semiconductor industry are looking beyond 14nm, however - and Intel is no exception. According to slides obtained by X-bit Labs, Intel has already begun research on sub-14nm processors with a view to getting 10nm parts into development for a 2015 roll-out.

According to the research and development pipeline espoused by the company's slides, 2015 will see the production of semiconductor components based on a 10nm process size, quickly followed by 7nm and 5nm parts.

The company has an uphill struggle ahead of it, however: the smaller the process, the larger the challenge. As the component sizes decrease and the gaps between components get smaller, numerous issues raise their heads. The biggest of these, current leakage, calls for a radical rethink to how semiconductors are designed in order to smash what has been termed the '10nm physical gate length barrier.'

Some companies have already developed prototype technology for the sub-10nm process, including IBM's carbon nanotube transitor, which is running in the lab based on a 9nm process size. Intel, however, isn't detailing the precise route its own research is taking, except to explain that it's in the process of researching new lithography, materials and interconnect techniques to address the issues of shrinking component sizes.

Intel's slide deck also explains that its 14nm production, due to begin in 2013 for the Broadwell family, will take place in its D1X Oregon, Fab 42 Arizona and Fab 24 Ireland facilities following 22nm upgrades to the D1D/C Oregon, Fab 32/12 Arizona and Fab 28 Israel plants.

Should Intel hit its schedule of a 10nm part by 2015, it will likely find itself ahead of its rivals in process technology - a move which will do nothing to lessen the company's growing dominance of the mainstream processor industry.

39 Comments

Discuss in the forums Reply
SpAceman 14th May 2012, 11:24 Quote
MOAR TRANSISTORS!

Doesn't seem that long ago that I was getting excited for 45nm... 65nm seemed small enough already at that point.
dunx 14th May 2012, 11:24 Quote
Or just MOAR thermal issues ?

dunx
SpAceman 14th May 2012, 11:31 Quote
Good point. Someone should tell Intel to stop following Moore's law. We don't want moar transistors.

I would like to think alternative semiconductor designs will deal with any thermal problems. Surely that is what they would be working on?
noizdaemon666 14th May 2012, 11:39 Quote
I beg to differ, we do want MOAR transistors, but we want them without needing LN2 to keep the CPU cool at stock speeds lol

I'm guessing they're already thinking about how to deal with the thermal issues given the poor thermal properties of Ivy Bridge.
mclean007 14th May 2012, 11:58 Quote
Quote:
Originally Posted by noizdaemon666
I'm guessing they're already thinking about how to deal with the thermal issues given the poor thermal properties of Ivy Bridge.
Care to clarify that? Thanks to 22nm and Tri-Gate, Ivy Bridge CPUs have a lower TDP than their Sandy Bridge predecessors. Specifically, both the 3770K and the 3570K have a TDP of 77W, against 95W for the 2700K and 2500K. The IB processors also apparently perform at least as well if not better in every situation at both stock and OC settings, have a faster GPU, and OC better. Doesn't sound like poor thermal performance to me.
wuyanxu 14th May 2012, 12:08 Quote
Quote:
Originally Posted by mclean007
Care to clarify that? Thanks to 22nm and Tri-Gate, Ivy Bridge CPUs have a lower TDP than their Sandy Bridge predecessors. Specifically, both the 3770K and the 3570K have a TDP of 77W, against 95W for the 2700K and 2500K. The IB processors also apparently perform at least as well if not better in every situation at both stock and OC settings, have a faster GPU, and OC better. Doesn't sound like poor thermal performance to me.
it's called Heat Density.

the CPU itself is great, performs well as you said. but the processor as a product is disappointing due to the poor thermal interface materials used between the die and the heatspreader. hence the "heat issue" mentioned.

solution is to change manufacturing process by soldering the heatspreader onto the die. at least that should work for a few years until 7nm made the "heat issue" resurface.
r3loaded 14th May 2012, 12:55 Quote
At this rate, Intel will need to work on their molecule-shrinking technology to keep Moore's law going.
rollo 14th May 2012, 13:04 Quote
both chips under stock settings stock cooler reach about a similar temparature

once overclocked is the temp dif

at 4.5ghz the temps are aproaching 100c on most poor air coolers really do need a decent chip + cooler to get a overclock going
Fat Tony 14th May 2012, 13:30 Quote
Looking forward to the first picometer chips
noizdaemon666 14th May 2012, 13:34 Quote
Quote:
Originally Posted by mclean007
Care to clarify that? Thanks to 22nm and Tri-Gate, Ivy Bridge CPUs have a lower TDP than their Sandy Bridge predecessors. Specifically, both the 3770K and the 3570K have a TDP of 77W, against 95W for the 2700K and 2500K. The IB processors also apparently perform at least as well if not better in every situation at both stock and OC settings, have a faster GPU, and OC better. Doesn't sound like poor thermal performance to me.
Quote:
Originally Posted by wuyanxu
it's called Heat Density.

the CPU itself is great, performs well as you said. but the processor as a product is disappointing due to the poor thermal interface materials used between the die and the heatspreader. hence the "heat issue" mentioned.

solution is to change manufacturing process by soldering the heatspreader onto the die. at least that should work for a few years until 7nm made the "heat issue" resurface.

As Wyx said to be honest. Just because they have a lower TDP doesn't mean they have better temperatures. As others have said OC temps are abysmal.
[USRF]Obiwan 14th May 2012, 13:39 Quote
Why not do a 32core at lower speeds with more cache. I really do not care if the processor takes 5x5cm or 1x1cm of space. In idle mode let 31 cores sleep to save loads of energy.
noizdaemon666 14th May 2012, 13:46 Quote
Quote:
Originally Posted by [USRF]Obiwan
Why not do a 32core at lower speeds with more cache. I really do not care if the processor takes 5x5cm or 1x1cm of space. In idle mode let 31 cores sleep to save loads of energy.

Pointless processor is pointless, at least in a consumer environment. You're hard pressed to find software that utilises anything more than 4 cores, let alone 32. In a server environment though, it'd be a good move.
dunx 14th May 2012, 14:09 Quote
I'm no expert in processor design, but how about spacing the four/six cores out on a bigger chunk of silicon, and adding the system RAM to fill the spaces in between ?

dunx
noizdaemon666 14th May 2012, 14:30 Quote
But then RAM wouldn't be upgradeable, at least not without removing the IHS.
jon 14th May 2012, 14:37 Quote
I, for one, welcome our sub molecular overlords ...
Jezcentral 14th May 2012, 15:24 Quote
When I see stuff like this, I begin to wonder if something DID crash at Roswell.
mecblade 14th May 2012, 15:31 Quote
Quote:
Originally Posted by noizdaemon666
I'm guessing they're already thinking about how to deal with the thermal issues given the poor thermal properties of Ivy Bridge.

The main reason for the poor thermal properties is the use of TIM (as Wuyanxu has stated) rather than Fluxless solder. A japanese review site managed to get a 14-24 degree Celsius reduction in temperature by using Coolaboratory TIM.

http://www.techpowerup.com/165882/TIM-is-Behind-Ivy-Bridge-Temperatures-After-All.html
mattbailey 14th May 2012, 15:33 Quote
Quote:
Originally Posted by Jezcentral
When I see stuff like this, I begin to wonder if something DID crash at Roswell.

Something did! Officially a weather balloon!
mclean007 14th May 2012, 15:45 Quote
Quote:
Originally Posted by wuyanxu

it's called Heat Density.

the CPU itself is great, performs well as you said. but the processor as a product is disappointing due to the poor thermal interface materials used between the die and the heatspreader. hence the "heat issue" mentioned.
It isn't really heat density. Heat density is a measure of thermal output per unit area / volume (depending on context). The issue you are describing is more about thermal conductance.

Anyway, nomenclature aside, I understand the point - IB chips get hotter than they would if the process was different - but the temperature in itself is pretty meaningless. What counts is how the chip performs, and (heat issue or no heat issue) IB chips are epic.
.//TuNdRa 14th May 2012, 16:28 Quote
But saying it like that is like saying "This car is amazing. it will explode and burst into flames if run for more than a certain time, but Amazing up until then." - Everything must be taken into account when saying things like that. Otherwise I could claim my Bulldozer is an amazing chip, crappy performance, and awful heat issues nonwithstanding, simply because it can mow-through 256bit decryption faster than Sandy Bridge processors.
MrGumby 14th May 2012, 16:32 Quote
IB cant be classed as epic. Surely that word implies a much bigger jump in performance. Maybe we can revist that word when Haswell shows up.
Jezcentral 14th May 2012, 16:54 Quote
Quote:
Originally Posted by mattbailey
Quote:
Originally Posted by Jezcentral
When I see stuff like this, I begin to wonder if something DID crash at Roswell.

Something did! Officially a weather balloon!
Of course!

*taps nose knowingly*
thehippoz 14th May 2012, 17:00 Quote
forget moores law.. murphys law and the snuffaluffagus is what I'm talking about
benji2412 14th May 2012, 17:23 Quote
Forgive me if this is wrong, but bearing in mind quantum tunneling occurs through barriers of 1.23nm, will this limit the size of current semiconductor technology?
noizdaemon666 14th May 2012, 17:26 Quote
Quote:
Originally Posted by mecblade
The main reason for the poor thermal properties is the use of TIM (as Wuyanxu has stated) rather than Fluxless solder. A japanese review site managed to get a 14-24 degree Celsius reduction in temperature by using Coolaboratory TIM.

http://www.techpowerup.com/165882/TIM-is-Behind-Ivy-Bridge-Temperatures-After-All.html

I know I read the article ;) That's still hotter than SB though.
Quote:
Originally Posted by mclean007
It isn't really heat density. Heat density is a measure of thermal output per unit area / volume (depending on context). The issue you are describing is more about thermal conductance.

Anyway, nomenclature aside, I understand the point - IB chips get hotter than they would if the process was different - but the temperature in itself is pretty meaningless. What counts is how the chip performs, and (heat issue or no heat issue) IB chips are epic.

SB was epic due to the massive increase in performance over Lynnfield and Nehalem. IB is not epic due to the heat increase and minimal performance increase.
John_T 14th May 2012, 17:33 Quote
Quote:
Originally Posted by mecblade


The main reason for the poor thermal properties is the use of TIM (as Wuyanxu has stated) rather than Fluxless solder. A japanese review site managed to get a 14-24 degree Celsius reduction in temperature by using Coolaboratory TIM.

http://www.techpowerup.com/165882/TIM-is-Behind-Ivy-Bridge-Temperatures-After-All.html

Now I, like most people on this site, am a fiddler by nature. I'll happily swap components in and out of my PC all day long. I'll swap CPU HSF's on and off & muck about experimenting with different TIMS - I'll change my GPU's HSF. No problem. I'll even, at a push, do a little light modding.

But bugger me: I absolutely draw the line at sawing up my brand new CPU to put better quality TIM INSIDE the processor case! Whatever Intel/AMD decide to do on that point, I'm afraid I for one am stuck with it... :)
John_T 14th May 2012, 17:33 Quote
Good luck to anyone who wants to give it crack though!
Beasteh 14th May 2012, 17:58 Quote
I remember the old Athlon XP chips that had no heatspreader... Crushing your core was a real concern. Good times?

By the sounds of things, Intel are sticking with silicon. It's a risky strategy: about 5 years ago it was said (and it's mentioned in the article, too) that 10nm was the limit, and we could go no further. It'll be interesting to see what Intel has up its sleeves to reach 5nm!
Showerhead 14th May 2012, 18:07 Quote
Quote:
Originally Posted by mecblade
The main reason for the poor thermal properties is the use of TIM (as Wuyanxu has stated) rather than Fluxless solder. A japanese review site managed to get a 14-24 degree Celsius reduction in temperature by using Coolaboratory TIM.

http://www.techpowerup.com/165882/TIM-is-Behind-Ivy-Bridge-Temperatures-After-All.html
What on earth are intel using as TIM that a simple change of compound gives that big a difference.
noizdaemon666 14th May 2012, 18:27 Quote
Quote:
Originally Posted by Showerhead
What on earth are intel using as TIM that a simple change of compound gives that big a difference.

Concrete ;)
Gradius 14th May 2012, 23:39 Quote
5nm! Holy cow!
Siwini 15th May 2012, 02:32 Quote
...and that's where 666 chips comes in...

Efficiency will improve and the size shrink, pretty soon they’ll get it down to where a human single organism can power up the processor.
fluxtatic 15th May 2012, 06:50 Quote
Sorry, but I have to take issue with this statement: "The majority of the industry is working on a 22nm process at present..." No, Intel is at 22nm. Nobody else is. Nobody in desktop, nobody in mobile. Maybe if you mean it as a bit of cheerleading in that Intel holds 90% or better of desktop marketshare, OK. But the statement itself (especially without clarifying "industry") is patently wrong.

IB isn't epic - SB was. I'm very curious to see how Haswell works out. Beyond that? Could be that this is just a bit of propaganda Intel let leak. They're at 22, and they're going to be at less than half that in three years? I've got my doubts.
Gareth Halfacree 15th May 2012, 07:43 Quote
Quote:
Originally Posted by fluxtatic
Sorry, but I have to take issue with this statement: "The majority of the industry is working on a 22nm process at present..." No, Intel is at 22nm. Nobody else is. Nobody in desktop, nobody in mobile. Maybe if you mean it as a bit of cheerleading in that Intel holds 90% or better of desktop marketshare, OK. But the statement itself (especially without clarifying "industry") is patently wrong.
Note "working on" - not "in production on," "working on." In other words (and if I'd been more careful with my phrasing) "The majority of the industry is working towards a 22nm process at present," except for Intel which is already there.
Elton 15th May 2012, 08:08 Quote
Why not make the die larger? Yeah that would reduce die farming and decrease efficiency, but surely more space to work on it (despite a smaller process) should be easier no?

Unless I'm simplifying it too much. Oh wait I am, there's also considerations of defective transistors in terms of density...
fluxtatic 15th May 2012, 08:53 Quote
Quote:
Originally Posted by Gareth Halfacree

Note "working on" - not "in production on," "working on." In other words (and if I'd been more careful with my phrasing) "The majority of the industry is working towards a 22nm process at present," except for Intel which is already there.

Fair enough. Can we chalk it up to you speaking "that funny English"? :P
Gareth Halfacree 15th May 2012, 09:05 Quote
Quote:
Originally Posted by fluxtatic
Fair enough. Can we chalk it up to you speaking "that funny English"? :P
In the words of noted philosopher Ralph Wiggum: "Me fail English? That's unpossible!"
wuyanxu 15th May 2012, 09:40 Quote
Quote:
Originally Posted by Elton
Why not make the die larger? Yeah that would reduce die farming and decrease efficiency, but surely more space to work on it (despite a smaller process) should be easier no?

Unless I'm simplifying it too much. Oh wait I am, there's also considerations of defective transistors in terms of density...
yes, larger die => less yield, more chance of defective die and thus a lot higher production cost.

from a business point of view, you'll want as small die as possible.
mdshann 17th May 2012, 22:21 Quote
Quote:
Originally Posted by Showerhead
Quote:
Originally Posted by mecblade
The main reason for the poor thermal properties is the use of TIM (as Wuyanxu has stated) rather than Fluxless solder. A japanese review site managed to get a 14-24 degree Celsius reduction in temperature by using Coolaboratory TIM.

http://www.techpowerup.com/165882/TIM-is-Behind-Ivy-Bridge-Temperatures-After-All.html
What on earth are intel using as TIM that a simple change of compound gives that big a difference.

I think it's funny that everyone is complaining about heat issues. Let's see what happens if we take a stock IB heatsink and use it with a 3 Ghz Prescott at stock speeds! Sure the stock temperatures haven't really changed, but the equipment needed to cool the chip to the same temperature has most certainly changed! Used to be you would need a stock 2.5" thick aluminium cooler with a copper core to reach temps of 50c to 60c. The CPUs out now will do better than that with an aluminium and copper cored cooler that is about 5/8" thick! Case in point, the computer I am on now is a Pentium D 2.8 Ghz at stock speed with a stock cooler, the side panel is not present, and it is running at 60c with just a web browser open to this page. The i3 next to me is running at about 45c idle with the stock 5/8" cooler. Where's this heat issue? These processors are designed to withstand up to 110c without damage. We're nowhere near that at the moment.

In my opinion, Intel is using a TIM because that's all that is needed. Sure, they could make the chips run cooler by using a different solution, but why would they? To a company like Intel, a few cents savings per chip will save them millions over the course of that chips manufacturing run time. In my experience TIM tends to last a good long time before it becomes an issue, maybe 8 or 10 years or so if the computer is kept relatively dust free. By that time, even your granny should be ready for a new computer.
Log in

You are not logged in, please login with your forum account below. If you don't already have an account please register to start contributing.



Discuss in the forums