Tuesday, March 7th 2017

Globalfoundries: 7 nm to Enable up to 2.7x Smaller Dies, 5 GHz CPUs

Globalfoundries' Chief Technical Officer, Gary Patton, talked about the future he believes can be possible in future manufacturing processes, calling for particular attention towards the next step in the ladder at 7 nm. Apparently, the 7 nm process at Globalfoundries has received a shot in the arm from the integration of ex IBM engineering specialists (remember that IBM practically paid Globalfoundries to take its manufacturing division of its hands), and the company now expects better than foreseen technical specs and achievements of its 7 nm process.

While a move from 14 nm to 7 nm was expected to provide, at the very best, a halving in the actual size of a chip manufactured in 7 nm compared to 14 nm, Gary Patton is now saying that the are should actually be reduced by up to 2.7 times the original size. To put that into perspective, AMD's 1000 series processors on the Zeppelin die and 14 nm process, which come in at 213 mm² for the full, 8-core design, could be brought down to just 80 mm² instead. AMD could potentially use up that extra die space to either build in some overprovisioning, should the process still be in its infancy and yields need a small boost; or cram it with double the amount of cores and other architectural improvements, and still have chips that are smaller than the original Zen dies.
According to Patton, these die space saving improvements aren't the only thing that has gone on better than they expected on the 7 nm manufacturing process. Patton said that he expects this design to be able to scale pretty well to some 5 GHz operating frequencies. Now, this is the least interesting part of the 7 nm equation, even though it might not seem like it. The ability to scale up to 5 GHz frequencies will of course depend on the architecture's design being able to achieve that operating frequency stably, most of all. And of course, we've already had an historical example of an architecture that aims to go as high as possible in the frequency department with Intel's NetBurst - and we all remember how that went.
It remains to be seen what these 7 nm expectations will mean for AMD, of course. But recent events have made it abundantly clear that the company is now in the more solid competition footing it ever has been with Intel when it comes to manufacturing processes. It seems that AMD's decision to spin off its manufacturing division was a heavy, but ultimately smart decision; let's see where this road will take us. Sources: AnandTech, via PC Games Hardware.de
Add your own comment

22 Comments on Globalfoundries: 7 nm to Enable up to 2.7x Smaller Dies, 5 GHz CPUs

#1
RejZoR
Problem with ever smaller dies is cooling. You got billions of transistors crammed in a tiny surface. And then you're expected to pull that heat efficiently through a small contact surface, preferably lubricated with a crappy TIM, to make things even worse...
Posted on Reply
#2
theoneandonlymrk
RejZoR said:
Problem with ever smaller dies is cooling. You got billions of transistors crammed in a tiny surface. And then you're expected to pull that heat efficiently through a small contact surface, preferably lubricated with a crappy TIM, to make things even worse...
Preferably lubricated with crappy Tim ???

This is Gf Pr what's team blues practices got to do with this , thats a confused statement you prefer crap tim ??

So your saying stick your improvements Gf then or am i miss reading your sentiment.

Seams odd i dunno.
Posted on Reply
#3
Tomgang
5 GHz dosent sounds to far out. I mean coffee lake can go to 4.7 GHz pretty easy and 5 GHz is not so rare either. Well time has to tell. I have even had my own old I7 980X close to 4.8 GHz on aircooling but at a pretty high voltage. So can we finally have a 5 GHz out of box cpu with out an tdp of 220 watt (yeah AMD i am thinking on the turd FX-9590, that had a max boost clock of 5 Ghz but whit a TDP of 220 watt it where to high a TDP for a stock cpu).

Until then i just have to hope the bedst and maybe finally i can use this avatar for good :peace:
Posted on Reply
#4
theoneandonlymrk
Tomgang said:
5 GHz dosent sounds to far out. I mean coffee lake can go to 4.7 GHz pretty easy and 5 GHz is not so rare either. Well time has to tell. I have even had my own old I7 980X close to 4.8 GHz on aircooling but at a pretty high voltage. So can we finally have a 5 GHz out of box cpu with out an tdp of 220 watt (yeah AMD i am thinking on the turd FX-9590, that had a max boost clock of 5 Ghz but whit a TDP of 220 watt it where to high a TDP for a stock cpu).

Until then i just have to hope the bedst and maybe finally i can use this avatar for good :peace:

You expecting intel to use glo fo???

Whats with the trolls today ,tut.
Posted on Reply
#5
trparky
Step 1: Buy a 2000-series Ryzen chip today.
Step 2: Wait a year and deal with a year of not-so-great 1080p gaming performance.
Step 3: Buy a 3000-series Ryzen chip when it comes out.
Step 4: ???
Step 5: Profit!!!
Posted on Reply
#6
Tomgang
theoneandonlymrk said:
You expecting intel to use glo fo???

Whats with the trolls today ,tut.
Lol. No trolling. Just hope that intel will finally have a cpu with a 5 ghz clock out of box. As far i know intel has never release a 5 ghz cpu before. They have only claimed that there devil canyon or i7 4790k cut oc to 5ghz, but turned out only a very few of then cut do it for real.
Posted on Reply
#8
RejZoR
theoneandonlymrk said:
Preferably lubricated with crappy Tim ???

This is Gf Pr what's team blues practices got to do with this , thats a confused statement you prefer crap tim ??

So your saying stick your improvements Gf then or am i miss reading your sentiment.

Seams odd i dunno.
When will people start understanding context and not just rush to dumb conclusions? If you'd actually done that you'd realize what I was saying was just shrinking things down doesn't solve all the problems. It actually creates problems and when company goes cheap it just makes things even worse.
Posted on Reply
#9
crazyeyesreaper
Chief Broken Rig
RejZoR said:
Problem with ever smaller dies is cooling. You got billions of transistors crammed in a tiny surface. And then you're expected to pull that heat efficiently through a small contact surface, preferably lubricated with a crappy TIM, to make things even worse...
While true keep in mind unlike Intel AMD is going the multiple chips per package. Meaning at the very least AMD in theory could shrink and simplify the design while using multiple smaller dies. Think threadripper using a similar strategy AMD could use less cores but more smaller dies under the IHS to better dissipate and keep heat per die under control. With the move away from monolithic designs there are more solutions available. While the package may be more expensive the ultimate cost per die comes down. As DDR4 and eventually DDR5 come around the speed and lower latencies improve how quickly these multiple dies can communicate. As such AMD can likely work around the issues a bit more easily than say Intel right now.

That said only time will truly tell which direction AMD actually takes.
Posted on Reply
#10
theoneandonlymrk
RejZoR said:
When will people start understanding context and not just rush to dumb conclusions? If you'd actually done that you'd realize what I was saying was just shrinking things down doesn't solve all the problems. It actually creates problems and when company goes cheap it just makes things even worse.
So make more sense in your first post and say what you mean then instead of mentioing tangential shit that is relevant to another company and foundrie and Not Gf.
You said some dumb shit im sorry for pulling on it though ok.
You think Anyone else thinks node shrinks are just That easy , you see they're doing it ,im sure as shit they know it's not that easy and have plans.
Posted on Reply
#11
john_
It's GlobalFailures. We know that we will be at least a little bit disappointed in the end.
Posted on Reply
#12
eidairaman1
The Exiled Airman
Tomgang said:
5 GHz dosent sounds to far out. I mean coffee lake can go to 4.7 GHz pretty easy and 5 GHz is not so rare either. Well time has to tell. I have even had my own old I7 980X close to 4.8 GHz on aircooling but at a pretty high voltage. So can we finally have a 5 GHz out of box cpu with out an tdp of 220 watt (yeah AMD i am thinking on the turd FX-9590, that had a max boost clock of 5 Ghz but whit a TDP of 220 watt it where to high a TDP for a stock cpu).

Until then i just have to hope the bedst and maybe finally i can use this avatar for good :peace:

8350, 5.0GHz across all 8 cores 24/7 operation, Air Cooled, 5GHz is old news to me.
Posted on Reply
#13
bug
RejZoR said:
Problem with ever smaller dies is cooling. You got billions of transistors crammed in a tiny surface. And then you're expected to pull that heat efficiently through a small contact surface, preferably lubricated with a crappy TIM, to make things even worse...
Nope, that's not the problem with smaller dies. When the transistors get smaller, so does the electrical charge they hold. The voltage needed to operate themgets lower and so does the heat. Roughly enough to compensate for the hight number of transistors. Desktop CPUs have remained in the 60-100W TDP since forever, surely you've noticed that. Surface area also didn't change much, even if at times some dies were smaller than others.
The problem with smaller transistors is actually the electrical charge. Leakage current that could have been acceptable at 16nm may very well you're loosing half the electrical charge you have left at 7nm.
Posted on Reply
#14
AsRock
TPU addict
RejZoR said:
Problem with ever smaller dies is cooling. You got billions of transistors crammed in a tiny surface. And then you're expected to pull that heat efficiently through a small contact surface, preferably lubricated with a crappy TIM, to make things even worse...
Yeah but at least AMD use solder, would be nice to see a break though cooling.

Tomgang said:
5 GHz dosent sounds to far out. I mean coffee lake can go to 4.7 GHz pretty easy and 5 GHz is not so rare either. Well time has to tell. I have even had my own old I7 980X close to 4.8 GHz on aircooling but at a pretty high voltage. So can we finally have a 5 GHz out of box cpu with out an tdp of 220 watt (yeah AMD i am thinking on the turd FX-9590, that had a max boost clock of 5 Ghz but whit a TDP of 220 watt it where to high a TDP for a stock cpu).

Until then i just have to hope the bedst and maybe finally i can use this avatar for good :peace:

TIM inside.
Posted on Reply
#15
theoneandonlymrk
eidairaman1 said:
8350, 5.0GHz across all 8 cores 24/7 operation, Air Cooled, 5GHz is old news to me.
+1
Posted on Reply
#16
ppn
More interested in 2-3nm that reduces the size of 213mm2 to 20mm2 or cram 10 times more cores. Vega 640 with 40960 processors and Zen 5++ with 80 cores. And hope they skip 5nm like they skip 10nm.
Posted on Reply
#17
Upgrayedd
RejZoR said:
When will people start understanding context and not just rush to dumb conclusions? If you'd actually done that you'd realize what I was saying was just shrinking things down doesn't solve all the problems. It actually creates problems and when company goes cheap it just makes things even worse.
I think what they meant was that Intel has their own fabs. So this has absolutely nothing to do with Intel or their TIM practices which might be crap. Then bug sorted out the whole shrinking sizes but same TDP statement.
Posted on Reply
#18
RejZoR
Upgrayedd said:
I think what they meant was that Intel has their own fabs. So this has absolutely nothing to do with Intel or their TIM practices which might be crap. Then bug sorted out the whole shrinking sizes but same TDP statement.
Context, you'e still not getting it. I never said ANYTHING about Intel, that's what you all instantly assumed. I literally just stated an universal fact that applies to ALL semiconductor chips. We're making them more complex and tinier. If we had TDP of 130W with chip the size of 20x20mm done on an old process in the past, it was much easier to remove heat from it thanks to massive contact surface than a modern 130W chip the size of 10x10mm. Again, before everyone freaks out and start jumping on my dimensions, this is a general example why there is a problem with shrinking of chips.
Posted on Reply
#19
EarthDog
You must take everyone around you for fools. Sure, you said that about die size and you are correct in a vacuum. Since amd uses solder this comment clearly infers intel and shows your feelings about its crappy paste youve mentioned in other threads.
...preferably lubricated with a crappy TIM, to make things even worse...
Posted on Reply
#20
medi01
If Global Foundries estimate to have 7nm mass production in "end of the year, most likely Q1 2019", how likely is AMD to already have working 7nm Vega chips?

trparky said:
Step 2: Wait a year and deal with a year of not-so-great 1080p gaming performance.
1) That "not so great 1080p gaming" with GTX 1080 is getting silly. Most of that 7.7% gap is a handful of games with spikes and and frame rates of 250+.
Show me the monitor that can dish that out.
2) Lower tier 2600x could fare much better vs comparably priced i5
Posted on Reply
#22
InVasMani
RejZoR said:
Context, you'e still not getting it. I never said ANYTHING about Intel, that's what you all instantly assumed. I literally just stated an universal fact that applies to ALL semiconductor chips. We're making them more complex and tinier. If we had TDP of 130W with chip the size of 20x20mm done on an old process in the past, it was much easier to remove heat from it thanks to massive contact surface than a modern 130W chip the size of 10x10mm. Again, before everyone freaks out and start jumping on my dimensions, this is a general example why there is a problem with shrinking of chips.
Can't they get around that issue a bit by going with multi socket boards thus increasing surface contact area for cooling? I get what you are saying at the same TDP heat transfer over time is harder on a smaller surface area. That clearly is a problem and one that isn't going away. At best we might come up with better cooling solutions that are hopefully economically viable.

I still really think that AMD should consider integrating a special tailored and kind of task specific single core CPU onto it's GPU's and/or chipsets as well. There is a lot of space on the backside of a GPU that could probably be utilized for such things. They could do the same with the chipset as well to help handle storage, network, ect related things. Why cram everything on same chip while adding heat to it as a byproduct on something that is shrinking and becoming more challenging to cool effectively at the same TDP levels? At some point or another chip makers are going to have to move more towards multi chip/socket solutions to these issues as we run into more challenging die node shrink limitations.

I hate to say it because I wish AMD luck in the future as they are a great company, but I strongly feel that Intel's FPGA is a enormous sleeping giant. FPGA's in general have so much potential to me as they can be configured appropriately to specific needs. I'm not sure why we don't have a CPU FPGA swamped with like 8 FPGA's around it that interconnect with it. You'd have tons of surface area for cooling and enormous amounts of reconfigurable power at hand especially if you had that with something like very potent APU at it's center with lots of AI machine learning capability to adapt to a users usage and needs.
Posted on Reply
Add your own comment