Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads

[Tigerick]

As Hot Chips 34 starting this week, Intel will unveil technical information of upcoming Meteor Lake (MTL) and Arrow Lake (ARL), new generation platform after Raptor Lake. Both MTL and ARL represent new direction which Intel will move to multiple chiplets and combine as one SoC platform.

MTL also represents new compute tile that based on Intel 4 process which is based on EUV lithography, a first from Intel. Intel expects to ship MTL mobile SoC in 2023.

ARL will come after MTL so Intel should be shipping it in 2024, that is what Intel roadmap is telling us. ARL compute tile will be manufactured by Intel 20A process, a first from Intel to use GAA transistors called RibbonFET.



Comparison of upcoming Intel's U-series CPU: Core Ultra 100U, Lunar Lake and Panther Lake

Model	Code-Name	Date	TDP	Node	Tiles	Main Tile	CPU	LP E-Core	LLC	GPU	Xe-cores
Core Ultra 100U	Meteor Lake	Q4 2023	15 - 57 W	Intel 4 + N5 + N6	4	tCPU	2P + 8E	2	12 MB	Intel Graphics	4
?	Lunar Lake	Q4 2024	17 - 30 W	N3B + N6	2	CPU + GPU & IMC	4P + 4E	0	12 MB	Arc	8
?	Panther Lake	Q1 2026 ?	?	Intel 18A + N3E	3	CPU + MC	4P + 8E	4	?	Arc	12

Comparison of die size of Each Tile of Meteor Lake, Arrow Lake, Lunar Lake and Panther Lake

	Meteor Lake	Arrow Lake (20A)	Arrow Lake (N3B)	Lunar Lake	Panther Lake
Platform	Mobile H/U Only	Desktop Only	Desktop & Mobile H&HX	Mobile U Only	Mobile H
Process Node	Intel 4	Intel 20A	TSMC N3B	TSMC N3B	Intel 18A
Date	Q4 2023	Q1 2025 ?	Desktop-Q4-2024 H&HX-Q1-2025	Q4 2024	Q1 2026 ?
Full Die	6P + 8P	6P + 8E ?	8P + 16E	4P + 4E	4P + 8E
LLC	24 MB	24 MB ?	36 MB ?	12 MB	?
tCPU	66.48
tGPU	44.45
SoC	96.77
IOE	44.45
Total	252.15

Intel Core Ultra 100 - Meteor Lake



As mentioned by Tomshardware, TSMC will manufacture the I/O, SoC, and GPU tiles. That means Intel will manufacture only the CPU and Foveros tiles. (Notably, Intel calls the I/O tile an 'I/O Expander,' hence the IOE moniker.)

 

[Fjodor2001]

So for Asus Zenbook S 14 (and similar premium low-power light-weight laptops), how much of the power consumption do we think is due to the display vs CPU when in low performance mode?

If the the display is what matters most, would it make sense to buy a variant with previous-generation CPU at lower cost?

 

[jdubs03]

More discussion about Royal Core:

https://www.reddit.com/r/intel/comments/1f945fl/some_rumors_about_the_royal_core_project

Make that of what you will.

 

[ondma]

More discussion about Royal Core:

https://www.reddit.com/r/intel/comments/1f945fl/some_rumors_about_the_royal_core_project

Make that of what you will.

I liked Pat as a choice initially. Now, I am not so sure. How many things can you cancel or sell off and still be a leading company?
Even Pat admitted everything hinges on 18A, and get the feeling that it is going to have problems as well.

As for Royal Core, maybe a low frequency, high IPC design would at least be very low power. Not sure how you can say it would have cost too much and had poor PPA without seeing a final product. Like I said in an earlier post, Intel is in a viscious spiral of needing a great product to turn them around, but dropping projects right and left. I dont see how they are going to turn things around like this. They are like a cancer patient who gets the worst possible prognosis at each step of diagnosis, while hoping desperately for a turn around.

 

[DrMrLordX]

Marketing speak aside, LNL is 10% faster than MTL @ 17W and 6% slower @ 23W. It loses steam fast, and that's with a node advantage. For the design target of 9W as declared by Intel the chip is very nice, but for 17W+ it could definitely use a bit more horsepower, either 4P6E or even 4P8E. That's why it's disappointing they're not following it up with another iteration.

Think of it this way, MTL loses to Hawk Point 8P/16T at 23W. LNL loses to MTL at 23W. We'll have independent numbers soon so I'd rather not try to extrapolate what happens at 17W, but it may not be the clear-cut win some were expecting.

That means Hawk Point will run the table on everything mobile Intel has @ 23W on up until Arrow Lake comes out (not counting high-wattage HX parts). And that doesn't even take Strix into account.

Oof if these are the yields for a 100mm2 chip on Intel 3

That would explain a few things about Sierra Forest. Doesn't bode well for Arrow Lake-U.

 

[Khato]

How many things can you cancel or sell off and still be a leading company?

With Intel? Plenty. Because in addition to stock buy backs and dividends buying up companies of potential interest/funding moonshot projects is all prior CEOs could think of to do with profits during the good years.

Royal had fanciful goals and hence it's of little surprise that there's still nothing to show after years of development.

 

[naukkis]

That means Hawk Point will run the table on everything mobile Intel has @ 23W on up until Arrow Lake comes out (not counting high-wattage HX parts). And that doesn't even take Strix into account.

You didn't notice that LL is thin&light form factor soc with 9-17w tdp? It design priority might not have been performance @23w tdp.

 

[DrMrLordX]

You didn't notice that LL is thin&light form factor soc with 9-17w tdp? It design priority might not have been performance @23w tdp.

Oh I noticed that, it's just that Intel has painted themselves into a corner.

 

[naukkis]

Oh I noticed that, it's just that Intel has painted themselves into a corner.

By releasing first true performant low&thin form factor soc, something that they could not do previously for Apple leading Apple to do it themself?

 

[511]

You didn't notice that LL is thin&light form factor soc with 9-17w tdp? It design priority might not have been performance @23w tdp.

Anything above is by OEMs request

 

[OriAr]

More discussion about Royal Core:

https://www.reddit.com/r/intel/comments/1f945fl/some_rumors_about_the_royal_core_project

Make that of what you will.

Sounds like a good riddance then.
While it was absolutely interesting from an engineering POV, 40% in ST perf over ADL with nearly double the die area per core would have been a very bad combination, especially since ARL-S looks to have roughly 25% better ST perf with the same die area already so it'd be double the area for a normal generational gain in ST perf, not exactly a great achievement.
Sometimes things get canned simply because they aren't as good as previously thought, looks like that's what happened here.

 

[LightningZ71]

If you go back a handful of years ago, Intel was likely thinking that they would have higher transistor densities available than they currently do. Those transistor counts wouldn't seem out of line in that world.

 

[IEC]

MLID (lol) believes Arrow Lake will end up just below the 7800X3D in the stack for gaming:


Only a little over a month until we can either roast him for being wrong or be disappointed that he was on the money.

 

[Kepler_L2]

MLID (lol) believes Arrow Lake will end up just below the 7800X3D in the stack for gaming:
View attachment 106942

Only a little over a month until we can either roast him for being wrong or be disappointed that he was on the money.

There's a good chance Arrow Lake ends up slower in gaming than Raptor Lake.

 

[9949asd]

MLID (lol) believes Arrow Lake will end up just below the 7800X3D in the stack for gaming:
View attachment 106942

Only a little over a month until we can either roast him for being wrong or be disappointed that he was on the money.

that’s MLID man lol.

 

[Josh128]

So for Asus Zenbook S 14 (and similar premium low-power light-weight laptops), how much of the power consumption do we think is due to the display vs CPU when in low performance mode?

If the the display is what matters most, would it make sense to buy a variant with previous-generation CPU at lower cost?

This is a very good representation of how much display consumes relative to CPU. Strix 370 gets 35% better battery runtimes going to 14" OLED from 16". It kind of wipes the floor with everything else. So yes, it makes a huge difference.

https://www.notebookcheck.net/Asus-VivoBook-S-14-OLED-laptop-review-Successful-performance-of-the-Ryzen-AI-9-HX-370.880476.0.html

 

[DokiDoki]

There's a good chance Arrow Lake ends up slower in gaming than Raptor Lake.

Ah yes, the same probability as the fabled "Zen 5 +40% over Zen 4 core-for-core".

 

[511]

This is a very good representation of how much display consumes relative to CPU. Strix 370 gets 35% better battery runtimes going to 14" OLED from 16". It kind of wipes the floor with everything else. So yes, it makes a huge difference.

https://www.notebookcheck.net/Asus-VivoBook-S-14-OLED-laptop-review-Successful-performance-of-the-Ryzen-AI-9-HX-370.880476.0.html

View attachment 106945

They test at 150 nits as well and the biggest problem what I find with oled is dark mode if someone uses light scheme he is bound to get drained quicker

 

[ondma]

Sounds like a good riddance then.
While it was absolutely interesting from an engineering POV, 40% in ST perf over ADL with nearly double the die area per core would have been a very bad combination, especially since ARL-S looks to have roughly 25% better ST perf with the same die area already so it'd be double the area for a normal generational gain in ST perf, not exactly a great achievement.
Sometimes things get canned simply because they aren't as good as previously thought, looks like that's what happened here.

Are you saying ARL will have 25% better ST performance than ADL? Pretty optimistic, to put it mildly.

 

[DokiDoki]

Are you saying ARL will have 25% better ST performance than ADL? Pretty optimistic, to put it mildly.

What's optimistic about ARL +25% over ADL?
RPL is already +15% over ADL (14900KS vs 12900KS).

 

[dttprofessor]

I am gonna call BS on that as there is no way Intel would have ramped up SRF and GNR if yields were that bad.
Raichu already said d0 for Intel 3 is somewhere in the 0.1-0.15 range.

IT‘s too good to believe.

 

[DrMrLordX]

By releasing first true performant low&thin form factor soc, something that they could not do previously for Apple leading Apple to do it themself?

By not releasing anything else.

 

[cannedlake240]

going to 14" OLED from 16"

*14" 1080p 60hz vs 16" 2.8K 120hz

 

[OriAr]

Are you saying ARL will have 25% better ST performance than ADL? Pretty optimistic, to put it mildly.

That's literally the boost according to GB6.
12900K gets around 2700, the leaked 285K results have it around 3400, 3400/2700 = 1.26, or 26% perf increase in ST.

 

[Josh128]

*14" 1080p 60hz vs 16" 2.8K 120hz

Point stands regardless, screen has much to do with battery life.

 

[jpiniero]

I am gonna call BS on that as there is no way Intel would have ramped up SRF and GNR if yields were that bad.

I've mentioned this before... I think they are using the YOLO strategy. The yields/defect rate probably doesn't matter as much as you think. It helps that the die should be very salvageable. Helps make some money off of the node.

At this point, any Xeon customers left are the types that aren't going to buy AMD even with AMD's products being so much better. So if Intel can't supply Granite, they'll just supply Sapphire or Emerald.