GlobalFoundries to skip 10nm and jump straight to 7nm

[Coaxialgamer]

Quote

Silicon fabrication company GlobalFoundries is reportedly planning to skip development of the 10 nanometer (nm) process, and is aiming to jump straight to 7 nm. The company currently operates a 14 nm FinFET node. In 2015 the company acquired semiconductor manufacturing assets from IBM, and is using them to fast-track its development. When it's ready, the 7 nm node will offer both optical and EUV (extreme ultra-violet) lithography. Driving the EUV product is an IBM 3300 EUV fabricator at the company's advanced patterning center, in its Albany, New York fab.

 

 

 

Just stumbled across this article , thought it was interesting . It's quite short though .I don't know when Gloflo plans to be ready with the process , but at least they seem to be advancing ...

 

https://www.techpowerup.com/225095/globalfoundries-to-skip-10-nm-and-jump-straight-to-7-nm  ( source ) 

[Bananasplit_00]

huh cant remember if any of the major manufacturers were planning something on the 10nm node, if not then sure why not but if they were GF are going to loose some market tbh

[themaniac]

good luck, even Intel and IBM are having a hard time with 7nm

[Kickflapper]

Very cool, however will the smaller node size equate to faster and better architecture and power usage now?

[Flibblebot]

Aren't Intel on 10nm? At least, I think they've just partnered with ARM and others to make more use of their 10nm fab?

 

I'm surprised GloFo is going to 7nm - I thought there were problems which meant it wasn't going to happen for another 18 months...?

[VerticalDiscussions]

Woo, they wanna speed up really quick!

[Coaxialgamer]

1 minute ago, Flibblebot said:

Aren't Intel on 10nm? At least, I think they've partnered with ARM and others to make use of their 10nm fab?

Their 10nm process won't be ready until late 2017. Intel's deal with arm and arm licensees only apply when the process is ready

[juretrn]

I thought the state of EUV was "almost, but not quite".

[Coaxialgamer]

Just now, juretrn said:

I thought the state of EUV was "almost, but not quite".

This will still take years before it is deployed 

[PRTI]

I wonder what will we use to get to 1nm,7nm is supposed to be released in 2018,they really wanna speed up the process of reaching Quantum Tunneling 

[themaniac]

1 minute ago, Nicholatian said:

GlobalFoundries acquired assets from IBM to make this a realistic proposition.

even then my point still stands both IBM and Intel are having a hard time getting to 7nm, even with IBM's assets its still going to be hard, iirc IBM has only been able to get a half chip(I think it a half chip or what ever they're called) so far, I don't see 7nm being ready for another 2-4 years

[Coaxialgamer]

5 minutes ago, keNNySOC said:

I wonder what will we use to get to 1nm,7nm is supposed to be released in 2018,they really wanna speed up the process of reaching Quantum Tunneling 

We will never get to 1nm,  at least not on silicon.  Silicon is expected to go down to 5-7nm max.  And since dennard scaling has broken down,  smaller sizes won't necessarily mean better power.  Plus,  the time and resources needed to design a chip with that process would be enormous.  Last time i checked,  a moderately complex chip took about 70M dollars and 20 man years ( divide that by the number of engineers on the job),  while a 14nm finfet chip of similar complexity costs ~250M dollars and takes around 500 man years. 

[themaniac]

Just now, Nicholatian said:

You chalk Intel and IBM up quite a bit, don’t you? They’re not perfect, and they have their weak points.

 

I don’t think GF would make this move if they weren’t planning to have 7nm ready to coincide with Intel’s 10nm or shortly thereafter.

well yeah they do have weakness's but they do have really skilled engineer's and the money to work on it

[PRTI]

Just now, Coaxialgamer said:

We will never get to 1nm,  at least not on silicon.  Silicon is expected to go down to 5-7nm max.  And since dennard scaling has broken down,  smaller sizes won't necessarily mean better power

We know that already,at 5nm its Quantum Tunneling which is used for Quantum Computers,and Quantum computers can't replace normal ones

[VerticalDiscussions]

1 minute ago, keNNySOC said:

-snip-

[Coaxialgamer]

Just now, Nicholatian said:

T R I G G E R E D

 

man years? man years!? man years!?!?!?!?!?!?!?!?!?!?!?!?!????? How dare you degrade women around the world with such a patriarchic label!!!!!!!!!!!

In all seriousness though,  it's man as in human 

[Curufinwe_wins]

1 minute ago, keNNySOC said:

We know that already,at 5nm its Quantum Tunneling which is used for Quantum Computers,and Quantum computers can't replace normal ones

LOL let's not talk about quantum effects if we don't know the qm...

 

Quantum tunneling is an issue for silicon even at 14nm, and is one of the biggest reasons Denard scaling has broken down. Additionally quantum computers rely on qubits, that is the primary difference.

[PRTI]

Just now, Curufinwe_wins said:

LOL let's not talk about quantum effects if we don't know the qm...

 

Quantum tunneling is an issue for silicon even at 14nm, and is one of the biggest reasons Denard scaling has broken down. Additionally quantum computers rely on qubits, that is the primary difference.

Quantum Computers can't be used for as normal ones as they offer allot of disadvantages,there BIG advantage is for cracking/solving really hard encryption's such as RSA4096

[Energycore]

Let's all be reasonable here and call them slave hours.

[Doobeedoo]

We're running out of nm's
As far as silicon being pushed I wonder what will be the last node, at what size and how long are going to stay on it until something entirely new comes. Like new material, 3D built CPUs and such. 

[PRTI]

2 minutes ago, Doobeedoo said:

We're running out of nm's
As far as silicon being pushed I wonder what will be the last node, at what size and how long are going to stay on it until something entirely new comes. Like new material, 3D built CPUs and such. 

Your right,we should go to pm or even better ym

[Curufinwe_wins]

18 minutes ago, gabrielcarvfer said:

Yes, but quantum tunneling can be used to improve things reducing power consumption and latency of transistors. \o/ 
http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7006647

In ideal cases sure.... but in the mean time almost any time you hear "electron leakage" or " bit rot" these aresults quanta effects that get exponentially worse with decreased pitch.

[riklaunim]

1 hour ago, Coaxialgamer said:

Their 10nm process won't be ready until late 2017. Intel's deal with arm and arm licensees only apply when the process is ready

I would say that they want to use those very small ARM chips as a meat shield to make the process moving with good yields instead doing it the hard way with their own bigger chips first.

[Coaxialgamer]

14 minutes ago, riklaunim said:

I would say that they want to use those very small ARM chips as a meat shield to make the process moving with good yields instead doing it the hard way with their own bigger chips first.

That's not exactly how it works.  Bigger chips yield differently than small chip,  and process maturity doesn't necessarily get better by running a lot of chips through it. 

[riklaunim]

They will likely have more defects at start of the new process than when they will get used to it and know it better.