Lakefield CD had to be that loosely packed to prove a point (of foveros actually working without catching fire)> Intel 7 is probably behind N4
> Intel 7 is probably around N5 but behind N5P
Uh, which one is it?
Anyway, I don't really mind the naming changes all that much. Although having to refer to 10ESF as 7nm now is just plain weird. And I don't expect Intel's 4nm to be measurably more dense than N5/N5P products, given how Intel tend to spam HP cells more than everyone else, with product densities tending to be smaller than the competition's (e.g. Lakefield compute tile is 49MTr/mm^2, despite being a die with no I/O)
Now all they need to do is rename 14nm to Intel 10 to maximize confusion.So they changed the strategy again and went with a Tick-Tock-Rename instead? /s
Intel also revealed the GPU EU count for Meteor Lake yesterday:
Up to 192 EUs
Ian saying Alder Lake is "now on 7nm" is confusing. It's still 10nmESF, just relabled as Intel 7...but throughout the article he kept saying ADLS is 7nm.
Now all they need to do is rename 14nm to Intel 10 to maximize confusion.
We were mocking the wrong freaking node all along!Now all they need to do is rename 14nm to Intel 10 to maximize confusion.
Intel is claiming a clear path to process performance leadership by 2025, but i don't see how that can be realistic
View attachment 47802
I mean they are behind both TSMC and Samsung in the race to GAA, and that is best case, if they can stick to their own roadmap this time.
View attachment 47803
He seems to be hedging the bet that 2023 is possible, but 2024 is the most reasonable timeline.
This would be a similar time frame to TSMC’s 2nm process node which will utilize gate all around transistor architecture, likely in 2024.
So it really seems like Intel is launching ADL-S at the end of October (as MLID claimed).
Rembrandt will definitely be on it's way out, 2023 is competing vs Phoenix.The thing with those 192EU, they are going to be fast, maybe even faster than RMB, but the Intel problem as always is timing, by the time Meteor Lake shows up RMB should be on its way out, if everything goes OK for both AMD and Intel. Meanwhile, we are going to have, on desktop, the biggest IGP gap in history the second RMB shows up.
What difference would standardized benchmarks make? Neither Intel's old names or the new names had any basis in physical reality, not for a long time.
When the node names did conform to physical reality everyone tried to scale by sqrt(2) every two years as that would double the number of transistors which fit Moore's Law and what ITRS dictated in its technology roadmap for semi equipment. They used various strategies to reduce leakage, increase drive current, etc. etc. that would allow higher clock rates but the process names were based solely on transistor density.
So if you want a "standardized benchmark" compare them by transistor density. Unless there is someone in this forum who actually makes decisions ordering wafers on a leading edge process, the node names only useful for Intel vs AMD vs Apple fanboying.
Rembrandt will definitely be on it's way out, 2023 is competing vs Phoenix.
That being said, Phoenix sticks to the same iGP - next improvement there is with Strix Point in 2024.
Nah, RMB will be fine. Sure Alder Lake will have a commanding lead for ST loads, and assuming it has the power it needs it'll beat RMB on MT too, but that's still a big if. We're talking about laptops here after all.A bit concerned for AMD in the mobile space, TBH. They seemingly have nothing to go up against Golden Cove as Rembrandt is "Zen 3+" (unless something is missing from the leaks). I guess we will see how it plays out. All I know is that in another 1-2 generations, many people won't need anything beyond a laptop if this progress keeps up.
i would say those are Consoles, and Intel has 0$ there.In the majority of devices where AMD currently finds themselves in now - the more reasonably sized gaming laptops - they'll have their niche where they'll be fine.
Meanwhile, we are going to have, on desktop, the biggest IGP gap in history the second RMB shows up.
A bit concerned for AMD in the mobile space, TBH. They seemingly have nothing to go up against Golden Cove as Rembrandt is "Zen 3+" (unless something is missing from the leaks). I guess we will see how it plays out. All I know is that in another 1-2 generations, many people won't need anything beyond a laptop if this progress keeps up.
Yeah I know that, but EUV mainly used for the BEOL? Shouldn't it be used mainly for the smaller constructs in the FEOL?
Yeah, i agree with ... Charlie of all people on this. They should have jumped directly to Angstrom with everything. There have been 130nm processes after all so even 70A would have been totally fine and would have avoided most of the backlash.Intel renaming their processes like that just seems stupid somehow. Now it's going to be difficult to understand what anyone has been saying about their 10nm and 7nm processes for, I don't know . . . years? Good job sowing confusion Intel.
Yeah, i agree with ... Charlie of all people on this. They should have jumped directly to Angstrom with everything. There have been 130nm processes after all so even 70A would have been totally fine and would have avoided most of the backlash.
It would have been even better to leave 10nm ESF as it is and rename the rest, but there is no need for this temporary "realignment" with samsung and TSMC nodes
You have to look at it this way: if Intel knocks it out of the park with ADL, the negative chatter about their fabs should die down.
Honestly, I don't care if Intel names their process nodes '10 flying butt monkeys'. If they get competitive again, great!. If they don't, rats!
14nm -> Intel 14
14nm+ -> Intel 13
14nm++ -> Intel 12
14nm+++ -> Intel 11
Before 2018, Skylake was 14nm until they changed it to 14nm+ for Cooperlake's release14nm -> Intel 14
14nm+ -> Intel 13
14nm++ -> Intel 12
14nm+++ -> Intel 11
...