Intel 7nm in 2021

[porina]

Quote

Intel provided an update regarding its upcoming fabrication technologies at its 2019 Investor Meeting. The company is on track to produce server-class products using its l0 nm manufacturing technology already in the first half of 2020. What is relatively surprising is that the company intends to start production of chips using its 7 nm process already in 2021.

Source: https://www.anandtech.com/show/14311/intel-process-technology-update-10nm-server-products-in-1h-2020-accelerated-7nm-in-2021

 

Not much info yet, as it seems the news is still coming in. We already knew they were planning to release in volume 10nm mobile parts this year, although there was no mention of desktop. This is indicating as comparable to TSMC 7nm. More interesting is Intel's 7nm claims. They had previously said the 10nm delays did not affect this as it was a separate path, and this is now indicating as due in 2021 to compete against TSMC 5nm. due to them specifying their process differently, the numbers are not directly comparable and we have this offset.

 

The impending release of Zen 2 based desktop CPUs on 7nm process will give AMD an advantage for at least the rest of the year from when it launches, and likely through 2020 also.

[The Benjamins]

aww cute, just in time to go againts TSMCs 5nm+ also planned for 2021.

 

https://wccftech.com/tsmcs-5nm-will-enter-mass-production-in-2021-claims-source/

 

 

[Bouzoo]

"World class packaging technology complements process leadership"

[Ravager911]

18 minutes ago, Bouzoo said:

"World class packaging technology complements process leadership"

"Lies! Deception!"

[porina]

19 minutes ago, Bouzoo said:

"World class packaging technology complements process leadership"

Now that you mention it, it sounds like random English words on a Japanese product used to make it look cool.

[.Apex.]

Is it confirmed that 10nm won't release on desktops this year? kinda weird to state that 10nm is coming this year when it's only for laptops, and they also mention servers

[Bouzoo]

5 minutes ago, Ravager911 said:

"Lies! Deception!"

[porina]

Just now, _Syn_ said:

Is it confirmed that 10nm won't release on desktops this year? kinda weird to state that 10nm is coming this year when it's only for laptops, and they also mention servers

They've said nothing about desktop parts that I'm aware of. Mobile parts for holiday 2019, so essentially year end. Server 1H next year. They're probably still going to be limited in what they can produce, and are targeting the biggest areas first.

[.Apex.]

1 minute ago, porina said:

They've said nothing about desktop parts that I'm aware of. Mobile parts for holiday 2019, so essentially year end. Server 1H next year. They're probably still going to be limited in what they can produce, and are targeting the biggest areas first.

Well hopefully they can still bring competition to AMD in those 2 years, it helps no one to favor one company over the other, I want both of them to innovate

[thorhammerz]

24 minutes ago, _Syn_ said:

Is it confirmed that 10nm won't release on desktops this year? kinda weird to state that 10nm is coming this year when it's only for laptops, and they also mention servers

10nm may not even make it to the desktop space - Intel had reportedly been dealing with issues of maintaining similar/greater clockspeeds as the 14nm+++ node (keeping in mind the clocks/IPC is the first and foremost metric for desktop chips, while power consumption is only important in the mobile/server space), along with possible issues shrinking the iGPU (Cannon lake 2c/4t chips couldn't even ship with a functional one).

 

Quite possible that Intel might jump directly to their 7nm node from 14nm for desktop chips, assuming they don't suffer another 5 years of setbacks.

[rawrdaysgoby]

So they are pretty much scrapping 10nm long term since they failed at it so badly and hopping straight towards 7nm since its development has been smoother then 10nm

[thorhammerz]

9 minutes ago, rawrdaysgoby said:

So they are pretty much scrapping 10nm long term since they failed at it so badly and hopping straight towards 7nm since its development has been smoother then 10nm

We'll know in a few years . Who knows what components might eventually end up being produced on that node.

 

Forum goers might ridicule Intel for their forward-looking marketing fluff on their approach on multi-chip(let) designs, but at the end of the day if it works then it works regardless of whether AMD / TSMC was "there" a year or two first.

[GoldenLag]

51 minutes ago, _Syn_ said:

Is it confirmed that 10nm won't release on desktops this year? kinda weird to state that 10nm is coming this year when it's only for laptops, and they also mention servers

considering we are getting 10 core 14nm CPUs

Spoiler

and this roadmap is on the table. its not looking good. 

[Techstorm970]

8 hours ago, GoldenLag said:

considering we are getting 10 core 14nm CPUs

  Hide contents

and this roadmap is on the table. its not looking good. 

Next spring, Intel is going to have 10 separate CPU architecture lines active at once!

 

Consumers:

[Guest]

The more + the more performance 

[Taf the Ghost]

Intel is really dressing up how bad this actually is.

 

Comet Lake (so Coffee Lake re-refresh) will go up to 10c, gets a new socket (so Z370/Z390 is a one-off) and will also cover all but the ultra-low power of the Mobile market.

 

Tigerlake will replace Icelake in the ultra-low mobile but it might also be the 28w mobile as well. Unknown if the 10nm++ will be as terrible at yielding, yet.

 

Rocket Lake will be Tigerlake back-ported to 14nm++. (That's the rumor at the moment.) 

 

Intel will move as fast as they can to 7nm, which is a EUV node for them. 2021 will probably be the early hyper-scaler Server Parts but consumer will still be mostly 14nm in 2021. The earliest you'll see Desktop 7nm should be 2022 called "Alder Lake".  Which will be going up against Zen4 CPUs from AMD. Intel has to be putting all of the resources they can into 7nm right now, as it's going to cost a lot to get it out early enough that they don't lose the Server Market starting in 2023.

[mr moose]

28 minutes ago, Taf the Ghost said:

 

Intel has to be putting all of the resources they can into 7nm right now, as it's going to cost a lot to get it out early enough that they don't lose the Server Market starting in 2023.

I doubt it,  developing node shrinks is not something you can really make go faster by throwing more cash at it.  Intel know this because their 10nm issues have highlighted the importance of getting it right rather than trying to be the first.  

[Taf the Ghost]

3 minutes ago, mr moose said:

I doubt it,  developing node shrinks is not something you can really make go faster by throwing more cash at it.  Intel know this because their 10nm issues have highlighted the importance of getting it right rather than trying to be the first.  

2022 is the really critical date for Intel. That's when the big Server vendors will start refreshing their "current building out" generation of servers products. (The ones they sell in big volume.)  If AMD can have a dominant product portfolio and 7nm is still running late, that's when Intel could get really hurt.

[mr moose]

Just now, Taf the Ghost said:

2022 is the really critical date for Intel. That's when the big Server vendors will start refreshing their "current building out" generation of servers products. (The ones they sell in big volume.)  If AMD can have a dominant product portfolio and 7nm is still running late, that's when Intel could get really hurt.

They have 7nm planed for then, rushing it will only serve to increase the chances of something going wrong.

[Taf the Ghost]

1 minute ago, mr moose said:

They have 7nm planed for then, rushing it will only serve to increase the chances of something going wrong.

I do agree that Intel actually needs to get 7nm yielding properly before rushing it out, though they're acting like it'll be fine. Though we can't ignore that pretty much every leading-edge node Fab that ran into problem doesn't get the next generation out properly on time, so this is going to be a long time to see how things go.

[bleedblue]

13 hours ago, Bouzoo said:

[Master Disaster]

I have a theory about Intels current situation, it's probably totally incorrect but meh, I'll share it anyway.

 

I'm not sure Intel are having trouble getting to 7nm, I think they've realised what they currently have is no match for what Zen will become and instead of focusing on only a node shrink, I think they're taking the time to do a full architecture redesign based around 7nm. Like AMD did when moving from FX to Zen, throw everything away and start again.

 

While they're doing it they can continue to improve the current architecture meaning unlike AMD they have competitive product on shelves and when they're ready they can drop the new line and retake the performance crown.

 

I'm struggling to believe that Intel are having so many problems getting to a node size others are already talking about surpassing, there's something else going on at blue camp.

[Stefan Payne]

13 hours ago, Ravager911 said:

"Lies! Deception!"

Lets just call it Corporate Bullshitting.

 

But yeah, don't look tooo good. What's Intel without the manufacturing advantage??

[GoldenLag]

16 minutes ago, Master Disaster said:

I think they're taking the time to do a full architecture redesign based around 7nm

based on what little we know regarding Intel`s roadmap i think we can assume that to be accurate. 

 

it now depends on how good that will be compared to what others have lined up at the time. Servers going directly for a first gen product could be risky, even if it is from Intel. 

[GoldenLag]

1 minute ago, Stefan Payne said:

What's Intel without the manufacturing advantage??

a refreshing can of a beverage you have been drinking since you were a child.

 

aka coke