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	8 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)	Arrow Lake Refresh (N3B)	Lunar Lake	Panther Lake
Platform	Mobile H/U Only	Desktop Only	Desktop & Mobile H&HX	Desktop Only	Mobile U Only	Mobile H
Process Node	Intel 4	Intel 20A	TSMC N3B	TSMC N3B	TSMC N3B	Intel 18A
Date	Q4 2023	Q1 2025 ?	Desktop-Q4-2024 H&HX-Q1-2025	Q4 2025 ?	Q4 2024	Q1 2026 ?
Full Die	6P + 8P	6P + 8E ?	8P + 16E	8P + 32E	4P + 4E	4P + 8E
LLC	24 MB	24 MB ?	36 MB ?	?	8 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.)

 

[igor_kavinski]

You literally posted it yourself one month ago



Look where the L3/LLC is

Thanks for pointing that out. So LNL is kind of a 5.5C/8T CPU (if 4 LP cores equal 1.5 P cores in MT performance <<< just a guess though).

 

[Hitman928]

You literally posted it yourself one month ago

Spoiler

Look where the L3/LLC is

Also I knew it even before that:

View attachment 90774

That's a simplified block diagram, not a layout. If Intel wants to say that the E cores starting on LNL are optimized for low power, that's fine, but that's not the same as what we were talking about. The discussion was about abandoning the P+E+LPE configuration to enable compute tile power off. LNL looks to abandon this in favor of a standard P+E configuration with all compute being back on a single die.

 

[SiliconFly]

Not sure why you think that quote is incongruent with what I am saying. The quote is stating where the LPE will be used with the current scheduler/firmware/thread director, then gives another option that if the LPE cores are to be used outside of those circumstances, the apps will need to be programmed specifically to use them. What do you see as incorrect?

Ok. I believe I'm not reading it right. Assuming I do, I think there's no way as of now to program a windows application or schedule a thread to specifically use the LP E cores alone.

 

[Hitman928]

Ok. I believe I'm not reading it right. Assuming I do, I think there's no way as of now to program a windows application or schedule a thread to specifically use the LP E cores alone.

I don't think anyone would do it to run solely on the LPE cores because programs shouldn't have the ability to initiate a turn off of the compute tile so it makes the effort unproductive, but you can write code to run on specific cores. I think his point was to be able to use the LPE cores a long with the P/E cores as desired by the programmer.

 

[Abwx]

You are spreading FUD unless you finally can prove your claim which you can't. Once again, after updating the bios Ultrabookreview got 1000 points more at 28W - there is no PL2 involved after 10 minutes. This score also is roughly in-line with the 28W score from Golden pig.

At Computerbase they show the CPU running at 35 or 45W even after 5mn, at this point it is in thermal equilibrium and wont get lower even after 10mn.

At NBC after the initial boost, in R15, the score decrease but not to the point that the CPU goes as low as 28W, rather something like 40-45W, at wich point the score is equivalent to about 13000-13500 pts in CB R23, at 28W it would hardly hit 12000 pts, so the 11700 score you were talking about as being dubbious is not that irrealistic.

So you believe GoldenPig doesn’t know how to limit power and chart a curve in 5W increments?
You also believe UltraBookReview is incapable of accurately measuring performance at a fixed power draw.

Methink that what he does is to select a given power and then bench with CB, but in all measurement points the turbo is engaged, so the score is partly done at high power, that s how he shaped the curve to mimick the one of RPL but at a higher efficency.

What happens if NoteBookCheck gets the same results with updated pcode? Will they also be incapable of measure a perf/watt curve?

You re talking out of thin air, they made enough measurements, among other that the laptop drain as much as 83-85W when launching Cinebench, from here we can use their CB R15 charts to estimate power after 10 runs.

Now if you think that a pcode can improve a process perf/watt overnight that s another discussion, for the time we have those numbers and they wont change.

 

[Khato]

The discussion was about abandoning the P+E+LPE configuration to enable compute tile power off. LNL looks to abandon this in favor of a standard P+E configuration with all compute being back on a single die.

Is LNL going to have E cores or LPE cores? In MTL the LPE cores reside within the SoC tile, but why can't the same advantages be present if they're in the CPU tile? The important distinction being that they're LPE cores using a low power interface to rest of system instead of the high performance ring bus used by standard E core.

I think a bit much has been made of the ability to turn off 'tiles' versus turning off logical blocks. As evidenced by AMD's designs, the important aspect in terms of efficiency is being able to turn off logical blocks on appropriate granularity.

 

[Abwx]

Sorry, but this statement is fundamentally flawed. It may be a gross approximation provided in entry level EE courses, but that doesn't make it correct.

Not at all, that s a physical law for mosfet, once you are above the threshold voltage Vth the mosfet conduction, and hence the switched current, increase as the square of the voltage, you can find the relation in wikipedia article on the subject, of course that s an idealised device with second orders effets not accounted, and those latter will worsen things actually.

Feel free to look at the V/F curve of any modern processor to disprove. eg the 12700k in this video -

- goes from .775V at 2GHz to 1V at 4Ghz, so about sqrt(1.66) voltage increase for 2X frequency. Meanwhile it goes from .825V at 2.5GHz to 1.35V at 5Ghz, which is about sqrt(2.67) voltage increase for 2X frequency.

I'm not going to attempt to explain in further detail than that though. While I recall the basics from an advanced semiconductor fundamentals course, since I don't work directly on process development or structural design I've long since forgotten the nuances necessary to provide a proper explanation.

ADL is not relevant because there s two kind of cores, the voltage is eventually overkill for one of the type of core, so at 775mV one already at a voltage that could allow higher frequency than 2GHz but the other type of core, or the uncore, would require such a voltage to work at a given frequency.

 

[controlflow]

So you believe GoldenPig doesn’t know how to limit power and chart a curve in 5W increments?
You also believe UltraBookReview is incapable of accurately measuring performance at a fixed power draw.

What happens if NoteBookCheck gets the same results with updated pcode? Will they also be incapable of measure a perf/watt curve?

He also claimed that the MTL 21 hour webpage refresh battery test result was "made up" by Hardware Canucks. If he doesn't like the result, the conclusion seems to be that it must be a fabrication, measurement error or a violation of the laws of physics.

 

[Hitman928]

Is LNL going to have E cores or LPE cores? In MTL the LPE cores reside within the SoC tile, but why can't the same advantages be present if they're in the CPU tile? The important distinction being that they're LPE cores using a low power interface to rest of system instead of the high performance ring bus used by standard E core.

I think a bit much has been made of the ability to turn off 'tiles' versus turning off logical blocks. As evidenced by AMD's designs, the important aspect in terms of efficiency is being able to turn off logical blocks on appropriate granularity.

I guess they could treat the E cores in LNL as basically an integrated island. Keeping them on a low power interconnect with the ability to turn off the P core island or E core island would help power but performance then with just 4 P cores or 4P + 4E over a slow interface could face some harsh performance penalties. AMD has really good granulation in terms of being able to control clock and voltage per core, but I don't know with what granularity they can turn off the cores/Vregs/clocks.

 

[H433x0n]

He also claimed that the MTL 21 hour webpage refresh battery test result was "made up" by Hardware Canucks. If he doesn't like the result, the conclusion seems to be that it must be a fabrication, measurement error or a violation of the laws of physics.

The results after the updated pcode aren’t even that crazy. It’s just the expected uplift from a new process.

 

[igor_kavinski]

So I'm wondering. Why isn't any website using Geekbench ML to compare the performance of NPUs in AMD/Intel/ARM CPUs?

Geekbench ML Results - Geekbench

 

[mikk]

At NBC after the initial boost, in R15, the score decrease but not to the point that the CPU goes as low as 28W, rather something like 40-45W, at wich point the score is equivalent to about 13000-13500 pts in CB R23, at 28W it would hardly hit 12000 pts, so the 11700 score you were talking about as being dubbious is not that irrealistic.

Hello? After 5-6 minutes the Asus goes down to 28W. The 10 min loop test is basically an 28W score. The best run would be PL2 affected and therefore scores quite a bit higher than the 10min loop run. You need to prove PL2 boosts higher.

For the 10 min loop, it starts at 50W and then gradually drops and stabilizes at around 28W (after about 5-6 minutes it gets to 28W).

Notebookcheck tested an Acer without the updated pcode. How is the Acer PL1 behavior relevant for the Asus? The Acer sustained PL1 might differ, it is a meaningless comparison.

You have no proof, nothing. Just baseless claims out of nothing. Also you can be sure when Golden pig makes a power curve it is what it says. You are basically saying he can't test properly.

 

[Geddagod]

You literally posted it yourself one month ago



Look where the L3/LLC is

Also I knew it even before that:View attachment 90774

Shhhh I forgot abt that, thank you for pointing out the sources lol
Wonder if this is dedicated Skymont LP E-cores, jus Skymont, or perhaps a port of crestmont on N3 (which would be incredibly dumb, but idk Intel does stuff sometimes lol).

 

[Hulk]

Thats transparently handled by the thread director. I don't think there's a way for users to configure how the cores are scheduled by the thread director as of now. But it'll definitely be a welcome addition. Someone should reach out to Intel.

Yeah, Intel has such an open door policy when it comes to taking suggestions I'm sure they'd love to hear from us.

 

[Geddagod]

He also claimed that the MTL 21 hour webpage refresh battery test result was "made up" by Hardware Canucks. If he doesn't like the result, the conclusion seems to be that it must be a fabrication, measurement error or a violation of the laws of physics.

I just tend to ignore him

 

[Geddagod]

Yeah, Intel has such an open door policy when it comes to taking suggestions I'm sure they'd love to hear from us.

their OEMs have quite an open door policy when it comes to leaking stuff haha

 

[FlameTail]

their OEMs have quite an open door policy when it comes to leaking stuff haha

The front door or the back one?

 

[Abwx]

Hello? After 5-6 minutes the Asus goes down to 28W. The 10 min loop test is basically an 28W score. The best run would be PL2 affected and therefore scores quite a bit higher than the 10min loop run. You need to prove PL2 boosts higher.




Notebookcheck tested an Acer without the updated pcode. How is the Acer PL1 behavior relevant for the Asus? The Acer sustained PL1 might differ, it is a meaningless comparison.

You have no proof, nothing. Just baseless claims out of nothing. Also you can be sure when Golden pig makes a power curve it is what it says. You are basically saying he can't test properly.

In their Cinebench R15 power measurement NBC found 83-85W for the whole laptop, and that s on an external screen, so the the OLED display is shut off, obviously MTL boost at 64W for the first run in this laptop.

The last runs have 14-15% lower score, this mean that power has tanked down from 60-64W or so to about 42-45W, certainly not 28W because at such a power the score would have tanked way more.

Now if you look at the same laptop tested by Computerbase their power curve start at 64W and then tank to 45W or 35W depending of the setting and dont go lower, one more time there s no power reduction down to 28W.

Actually you are relying on a dubbious reviewer because his numbers please you more than the ones of NBC and Computerbase, he didnt even make power measurements to prove his numbers, not even a basic watt metre connected in the main, that s all talks that we are asked to take at face value.

https://www.notebookcheck.net/Acer-Swift-Go-14-review-The-Meteor-Lake-Core-Ultra-7-impresses-with-its-AI-core.783349.0.html

Intel Meteor Lake anhand des Core Ultra 7 155H im Test

Das Acer Swift Go 14 ermöglicht einen Ersteindruck zu Meteor Lake, der dieselbe Single-, mehr Multi-Core- und viel mehr GPU-Leistung bietet.

www.computerbase.de

 

[mikk]

Now if you look at the same laptop tested by Computerbase their power curve start at 64W and then tank to 45W or 35W depending of the setting and dont go lower, one more time there s no power reduction down to 28W.

Neither Golden Pig nor Ultrabookreview tested with it. The Asus Zenbook goes down to 28W confirmed by the tester.

As for the Acer and 28W, Notebookcheck tested with a fixed 28W and not Computerbase. It's a fixed power usage at 28W and 35W and 45W. 10573 points at 28W which is quite a bit below the updated pcode scores.

We still evaluated the multi-core performance of the new Core Ultra 7 155H at fixed power limits of 28 Watts, 35 Watts, and 45 Watts in Cinebench R23.

 

[TESKATLIPOKA]

Notebookcheck will test a laptop where they will also include the power consumption during CB look. Just not sure when, It's dependent on who is doing the review.

If you can buy a MTL or PHX for a good price, you won't make a mistake, there aren't that big differences between them.
Maybe only later you or me will regret It, when Strix is out, but who knows when and at what price they will be released.

 

[FlameTail]

Damn. I now have a fancy for Zenbooks.

The OLED screen is beautiful and the all metal build quality is rare under $1000.

 

[Abwx]

Neither Golden Pig nor Ultrabookreview tested with it. The Asus Zenbook goes down to 28W confirmed by the tester.

As for the Acer and 28W, Notebookcheck tested with a fixed 28W and not Computerbase. It's a fixed power usage at 28W and 35W and 45W. 10573 points at 28W which is quite a bit below the updated pcode scores.

NBC measured power at the wall with the internal screen disabled, the laptop hoover at 83-85W at the start of Cinebench, you think that a laptop that pull this amount of power has a CPU that run at a fixed 28W..?..

Look better, they specify 45W sustained in the beginning of the article, more explicitely 55W PL2 and 45W PL1, and that s after they asked Acer to reduce PL2 down to 55W from the stock 64W wich is eventually still enabled.

55 W PL2 / Short Burst, 45 W PL1 / Sustained, Meteor Lake-H

After a lively exchange with Acer, we decided to remove the software. We were also able to lower the PL2 to 55 watts while leaving the PL1 at 45 watts, all with the aid of TechPowerUp's Throttle Stop

https://www.notebookcheck.net/Acer-Swift-Go-14-review-The-Meteor-Lake-Core-Ultra-7-impresses-with-its-AI-core.783349.0.html

 

[TESKATLIPOKA]

Damn. I now have a fancy for Zenbooks.

The OLED screen is beautiful and the all metal build quality is rare under $1000.

Zenbooks are using only one fan If I am not mistaken.
Check out If Ideapad 5 Pro is not available under $1000 with MTL. I have the 14-inch one with PHX and is very good, all metal, 2 fans, one extra USB-A port, OLED display. The only thing better at Zenbook should be the integrated numpad in touchpad.

p.s. Ideapad also has a privacy shutter, not sure about Zenbook.

 

[TESKATLIPOKA]

Doublepost.

 

[coercitiv]

Zenbooks are using only one fan If I am not mistaken.
Check out If Ideapad 5 Pro is not available under $1000 with MTL. I have the 14-inch one with PHX and is very good, all metal, 2 fans, one extra USB-A port, OLED display. The only thing better at Zenbook should be the integrated numpad in touchpad.

And no Armoury Crate. Lenovo Vantage is optional, whereas AC finds a way into your system.