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.)

 

[jdubs03]

It will not cross 170 cause the geekbench inflation is due to the Matrix extension it will be around 165

Fair enough. Using Geekbench 5 where that additional bump due to SME isn’t utilized in the benchmark, the improvement is ~16.75%, so going by that and the correlation remains it could breach 165 maybe up to 167.

 

[511]

Looks like the score is improving nicely for LNL

LENOVO 21NSSIT021 - Geekbench

Benchmark results for a LENOVO 21NSSIT021 with an Intel Core Ultra 7 256V 2.20GHz processor.

browser.geekbench.com

 

[SiliconFly]

Thats interesting. If LNL is posting such good ST scores, ARL-S might actually beat the competition in ST (and obviously MT too).

 

[Magio]

Thats interesting. If LNL is posting such good ST scores, ARL-S might actually beat the competition in ST (and obviously MT too).

256V is the lowest tier Ultra 7 SKU, too. It only boosts to 4.8GHz (confirmed by the Geekbench results page) while 266/8V go up to 5.0. Not a massive difference but not insignificant either. (The 30W PL1 chip goes up to 5.1 but it's not really comparable.)

 

[AMDK11]

View attachment 103518
Looks like the score is improving nicely for LNL

LENOVO 21NSSIT021 - Geekbench

Benchmark results for a LENOVO 21NSSIT021 with an Intel Core Ultra 7 256V 2.20GHz processor.

browser.geekbench.com

I wonder what the clock speed was in ST mode.

 

[511]

I wonder what the clock speed was in ST mode.

 

[AMDK11]

That is 0.9-1.0GHz less than the flagship ArrowLake-S.

Is there any possibility of comparison with Zen5?

 

[SiliconFly]

That is 0.9-1.0GHz less than the flagship ArrowLake-S.

Is there any possibility of comparison with Zen5?

An excerpt from TweakTown:

"The new Ryzen AI 9 HX 370 "Strix Point" APU scored 2010 points in Cinebench R23's single-core test and 23302 points in the multi-core test. The single-core result beats Intel's current flagship mobile CPU, the Core Ultra 9 185 "Meteor Lake" CPU (which scores 1935 points), and just 17147 in the multi-core run against 23302 from the Ryzen AI 9 HX 370.

Intel still has the lead with its desktop-class 55W version of the Core i9-14900HX, which scores 2193 points in the single-core run of Cinebench R23, and 30411 points in the multi-core run. AMD's own Ryzen 9 7945HX3D processor also beats the new Zen 5-based Ryzen AI 9 HX 370 "Strix Point" APU, but not in single core (the 7945X3D scores 1935 points in single-core) but beats them all in multi-core scores with 32782 points.

Read more: https://www.tweaktown.com/news/9940...re-ultra-185h-in-cinebench-r23-run/index.html"

Looking at this, it appears ARL might match or beat the competition in ST.

 

[AMDK11]

An excerpt from TweakTown:

"The new Ryzen AI 9 HX 370 "Strix Point" APU scored 2010 points in Cinebench R23's single-core test and 23302 points in the multi-core test. The single-core result beats Intel's current flagship mobile CPU, the Core Ultra 9 185 "Meteor Lake" CPU (which scores 1935 points), and just 17147 in the multi-core run against 23302 from the Ryzen AI 9 HX 370.

Intel still has the lead with its desktop-class 55W version of the Core i9-14900HX, which scores 2193 points in the single-core run of Cinebench R23, and 30411 points in the multi-core run. AMD's own Ryzen 9 7945HX3D processor also beats the new Zen 5-based Ryzen AI 9 HX 370 "Strix Point" APU, but not in single core (the 7945X3D scores 1935 points in single-core) but beats them all in multi-core scores with 32782 points.

Read more: https://www.tweaktown.com/news/9940...re-ultra-185h-in-cinebench-r23-run/index.html"

Looking at this, it appears ARL might match or beat the competition in ST.

Geekbench 6.1.0 for Windows AVX2?

 

[controlflow]

I expect LNL will be the closest thing to the Apple M series but for Windows. I really think it will make the current Snapdragon X stuff largely irrelevant since LNL will not have compatibility issues and will have a better NPU and a massively better GPU. I think it is likely that it will finally put to rest this notion that ARM ISA magically has some kind of significant low power advantage over x86. It really just comes down to core microarchitecture, system architecture and strong software/OS integration to do low power properly. LNL should be much improved in all of those areas and is probably the first real sane low power effort from the x86 camp.

 

[SiliconFly]

I expect LNL will be the closest thing to the Apple M series but for Windows. I really think it will make the current Snapdragon X stuff largely irrelevant since LNL will not have compatibility issues and will have a better NPU and a massively better GPU. I think it is likely that it will finally put to rest this notion that ARM ISA magically has some kind of significant low power advantage over x86. It really just comes down to core microarchitecture, system architecture and strong software/OS integration to do low power properly. LNL should be much improved in all of those areas and is probably the first real sane low power effort from the x86 camp.

Very true!

 

[Ghostsonplanets]

Was this posted here?:

https://twitter.com/x/status/1814496922574479504

Wildcat Lake is a low-cost SoC line-up that is meant to slot in products between ADL-N and MTL/ARL/PTL-U series (So, basically, budget laptops). It will feature P, E and LPE cores of unknown uArchs (Raptor/Grace? Redwood/Crest? Lion/Sky?) and also will feature a NPU based on Meteor Lake NPU (Or so was the original plan of the project).

IMO It's very likely to be an optimization/waterfall version of Meteor Lake with smaller ACM+ GPU and reduced IO/SoC (Fused tiles?).

 

[inf64]

I'm not sure if this was posted before: https://news.mynavi.jp/article/20240629-2975570/

It's a series of articles (3 parts) detailing LNL architecture. It's in JP so translator is needed (unless you know Japanese ).

 

[DavidC1]

I'm not sure if this was posted before: https://news.mynavi.jp/article/20240629-2975570/

It's a series of articles (3 parts) detailing LNL architecture. It's in JP so translator is needed (unless you know Japanese ).

Based on translate the guy is saying the 2x3 cluster on Gracemont was used for Hyperthreading. I stopped reading. When "technical" writers lack fundamental understanding of what they are writing about, it's worth less than toilet paper after being used for it's intended purpose.

🤦‍♂️

 

[Geddagod]

 

[coercitiv]

@Geddagod AFAIK Sierra Forest does not have SMT.

Why do I get the feeling that this article was "generated" with AI? [Edit:] Yeah, the link about the 576 threads leads to a slide from Intel's deck... showcasing a dual socket Sierra Forest.

Spoiler

They are either complete amateurs or using LLM to generate articles, which is basically the same these days.

 

[511]

Af

@Geddagod AFAIK Sierra Forest does not have SMT.

Why do I get the feeling that this article was "generated" with AI? [Edit:] Yeah, the link about the 576 threads leads to a slide from Intel's deck... showcasing a dual socket Sierra Forest.

Spoiler

View attachment 103600

They are either complete amateurs or using LLM to generate articles, which is basically the same these days.

Afaik the 576 thread version had dual 288 cores SRF which is Insane NGL

 

[Saylick]

Crikey, it doesn’t look good.

https://twitter.com/x/status/1816099937500684755

For those who don’t have Xitter:

 

[cebri1]

A 17% increase in CB23 from the 14900K doesn't look good? What I'm missing here.

edit: nevermind

 

[SiliconFly]

Crikey, it doesn’t look good.

Thats tragic! ARL-S complete washout in ST?

 

[Henry swagger]

Crikey, it doesn’t look good.

https://twitter.com/x/status/1816099937500684755

For those who don’t have Xitter:
View attachment 103695

Stil a qs sample.. tsmc n3B doesn't look efficient yet

 

[Saylick]

A 17% increase in CB23 from the 14900K doesn't look good? What I'm missing here.

I think the 14900K does a little better at CB23 than that Xitter post shows. For example, from TechPowerUp:


Idk how AT’s sample scored so high either:


From Tweaktown:

 

[inf64]

I'm not sure which profile was used for 14900K but according to CPU monkey, the SKU should score 38700 pts.
For reference 7950x gets 38.6K , while unofficially the 9950X gets 43.9K pts.

 

[cebri1]

I think the 14900K does a little better at CB23 than that Xitter post shows. For example, from TechPowerUp:

nah, disregard my post, the issue is with the ST performance. Also MT is behind expectation, still a QS sample tho.

 

[511]

nah, disregard my post, the issue is with the ST performance. Also MT is behind expectation, still a QS sample tho.

QS is closest thing to retail the issue is the Single Thread