Intel Might Drop 10nm Node for Desktop Processors, 14nm Until 2022?

[GoldenLag]

2 hours ago, sowon said:

I like the image that Intel is like the Internet Explorer of the CPU space in this context regarding shrinks.

Dont worry, Intel will soon come out with the cutting Edge

 

not sorry

[mr moose]

I wonder what 1 year of sales is worth to Intel in terms of R+D and new products?

[Harry Voyager]

I'm reminded of the period in the early 2000's when the Pentium 4 came out and faltered. I don't believe Intel fully got back in the game until Core 2.

 

I've been going back and forth on what I want to use for my rebuild, and with the likelyhood of a 5ghz Zen 3 AM4 part next year, and the likely delays in a desktop grade Ice Lake, I'm leaning more and more towards the AMD side right now, even with a very single thread heavy primary application. 

[melete]

5 hours ago, bcredeur97 said:

Intel has very tight tolerances on their processes I think that's why 10nm was such a struggle for them.

However, if they can't do 10nm... then why could they do 7??? that doesn't make sense to me.

Different processes, different teams, and Intel can apply the lessons it's learned from 14nm and 10nm towards 7nm. They were a little over-ambitious as to what they hoped to achieve with DUVL on 10nm, but Intel sounds pretty confident about what it can achieve with its EUVL 7nm process. If 10nm was a little too aggressive about what it could achieve with regards to scaling targets, 7nm sounds like it will be a more conservative process with less usage of multipatterning.

 

This is also why the "no 10nm desktop" stuff makes quite a bit of sense here. 7nm is on-time for 2022, so if Intel can't get 10nm out before say late 2021, that's going to be a very short-lived product lineup.

[Fnige]

Catchup Time

[LAwLz]

Friendly reminder that you can not compare the # nm from one plant to another. 

Intel 10nm is very different from TSM 10nm for example. 

 

Also a friendly reminder that you can not use two different processors with different architectures to compare the lithography. If you do then you're changing too many variables to accurately narrow down what is actually causing the difference in thermals, clocks, IPC, latency, power consumption, etc. 

[leadeater]

7 hours ago, melete said:

This is also why the "no 10nm desktop" stuff makes quite a bit of sense here. 7nm is on-time for 2022, so if Intel can't get 10nm out before say late 2021, that's going to be a very short-lived product lineup.

Ice Lake-SP is still targeted to release H1 2020 using 10nm+ and are server parts. With the supply issues Intel is having for 14nm parts, which I think is due to the 10nm delay because fabs that were supposed to be operational are not, if 10nm desktop is not to be a thing I think that has more to do with moving supply constraint off 14nm by way of server parts on 10nm+ which that fab process won't be able to supply multiple market segments.

 

By the time Intel has enough capacity in 10nm+ 7nm will already be commercially viable so unless there is a benefit 10nm+ that 7nm does not have, high clocks or something, there isn't any point using it.

[GoldenLag]

8 hours ago, Harry Voyager said:

5ghz Zen 3 AM4

unlikely at this point. 

 

new node isnt expecting any real change in clocks iirc. its mostly a yields improver and density improver. 

[Harry Voyager]

7 hours ago, GoldenLag said:

unlikely at this point. 

 

new node isnt expecting any real change in clocks iirc. its mostly a yields improver and density improver. 

I'll have to check, but I recall redgaming had a note up that the new node is expected to deliver about an 8% clock boost, along with the IPC increase. If AMD is able to resolve why they aren't getting the 4.7Ghz boost speed they've been targetting with the 3950X then a 5Ghz Zen 3 should be doable.

 

If all Intel is able to counter with is a 5.1Ghz 10th gen 14nm, it's going to be ugly for them until they get the new node up and running. 

 

Though, as the Spartan said, "if". We will see. 

[porina]

12 minutes ago, Harry Voyager said:

I'll have to check, but I recall redgaming had a note up that the new node is expected to deliver about an 8% clock boost, along with the IPC increase.

From memory 8% was the claimed IPC increase, clocks I think up to 200 MHz increase was thrown around. Still, the two combined is a nice boost if it happens.

[Unixsystem]

On 10/15/2019 at 8:19 AM, Taf the Ghost said:

Except it really isn't. Intel had to dial back 10nm really far to get it out the door. Far enough that no one seems to know what the node even is anymore. It's a shrink from their 14nm, but TSMC/Samsung should have a significant density advantage in actual shipping silicon.

 

Edit: It'd probably be safe to call Intel's 10nm as equal to something like the Foundry 9nm. Denser than Foundry 10nm but not as dense as Foundry 7nm.

This is my problem with a lot of people, especially on more knowledgeable forums like this, who still repeat the "Intel nodes are always better than the competitors'" line.  Historically it's certainly been true, but right now we have ample evidence that 10nm has been nerfed and no real solid numbers to know by how much. Speculating about the potential of their original goals can be fun, but if it was completely impossible/impractical to produce then it's basically just vaporware. 

 

I hope 7nm is much smoother sailing for them and it winds up being a generational leap over even the original Intel 10nm goals, but at this point they have had so many setbacks that I'm just not willing to give them a pass and say "oh 7nm is on track and will match Samsung/TSMC 5nm/5nm+/whatever". I think some healthy scepticism over their manufacturing upgrades is well earned by this point. 

[Belgarathian]

BRB, investing my life savings in AMD. 

[Fnige]

4 minutes ago, Belgarathian said:

BRB, investing my life savings in AMD. 

dont give up this early

[pizapower]

The 7nm EMIB would be pretty cool.

[leadeater]

6 hours ago, Waffles13 said:

but right now we have ample evidence that 10nm has been nerfed and no real solid numbers to know by how much

Intel did actually state what the original 10nm targets were and the revised ones now, don't remember what the figures are but they exist if someone wants to find them. Intel 10nm is still pretty darn good but it's also a great example of why EUV needs to exist (sooner).

[leadeater]

https://www.techcenturion.com/7nm-10nm-14nm-fabrication

 

Note: Maximums possible, does not mean actual used in specific products etc.

 

 

https://fuse.wikichip.org/news/2004/iedm-2018-intels-10nm-standard-cell-library-and-power-delivery/4/

[pizapower]

We need Intel. There wouldn't be AMD without Intel.

[pizapower]

It's like BMW vs MB, but BMW is better than Mercedes because BMW makes Rolls-Royce cars.

[pizapower]

So who is BMW and who is Mercedes?

[ImNotGayBut5BuckIs5Bucks]

the excalibur . now thats a nice ride to make BMW or MB jelly

[pizapower]

7 minutes ago, ImNotGayBut5BuckIs5Bucks said:

the excalibur . now thats a nice ride to make BMW or MB jelly

 The G63 AMG is better
Edit: My bad I thought you talk about the Ford Expedition

[pizapower]

Seems like that Excalibur is based on a MB. So it can't make Mercedes jelly about that car.

[Unixsystem]

19 hours ago, leadeater said:

Intel did actually state what the original 10nm targets were and the revised ones now, don't remember what the figures are but they exist if someone wants to find them. Intel 10nm is still pretty darn good but it's also a great example of why EUV needs to exist (sooner).

This is kind of what I'm talking about, though. The original plans looked like an incredible leap, but then it turned out that it was for all practical purposes impossible to actually produce, and so it got nerfed. 

 

Now we have the new nerfed 10nm, which on paper may still be way better than the competition, but they still seem to be having significant trouble bringing it to market in any meaningful capacity; this very topic is about the fact that it seems likely they will just skip it entirely for a significant percentage of their product line. 

 

That doesn't mean it isn't still a solid node in terms of theoretical performance, but if they can't translate that into shipping products then I have trouble giving them too much credit for having a  "darn good node" overall. Cool that they are trying to push the boundaries, but putting all their eggs into that one basket seems like it could go down as one of the biggest fumbles in the history of this industry. 

 

That said, if they do in fact get 7nm out in a reasonable time with the performance that you would hope for, then 10nm will probably go down as little more than a footnote (one which strangely enough will be more relevant to AMD than Intel).

[RoseLuck462]

Well for the time being I'm only interested in Ryzen Lake CPUs.

[Results45]

.

On 10/15/2019 at 5:46 AM, BiG StroOnZ said:

 

On 10/15/2019 at 8:53 AM, mcbaes72 said:

Intel responded to rumors... 

 

Intel: "Yes, There Will be 10nm Desktop CPUs"

 

My guess is that we'll at least see some form of Tiger Lake.