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.)

 

[Henry swagger]

Intel Core Ultra 9 285K "Arrow Lake" Desktop CPU Rumored To Clock Around 5.5 GHz

Intel's Arrow Lake CPUs such as the flagship Core Ultra 9 285K are rumored to feature a clock speed of up to 5.5 GHz.

wccftech.com

20A is not clocking as high as the 14900ks. Its only at alder lake clocks.. arrow lake 12% higher single core than 14900ks with clock regression 🤔🤨.. zen 5 will have the clock advantage witj n4p 🤨

 

[uzzi38]

Intel Core Ultra 9 285K "Arrow Lake" Desktop CPU Rumored To Clock Around 5.5 GHz

Intel's Arrow Lake CPUs such as the flagship Core Ultra 9 285K are rumored to feature a clock speed of up to 5.5 GHz.

wccftech.com

20A is not clocking as high as the 14900ks. Its only at alder lake clocks.. arrow lake 12% higher single core than 14900ks with clock regression 🤔🤨.. zen 5 will have the clock advantage witj n4p 🤨

Reading through the post, it looks like MebiuW actually said the following:

14900K should clock ~12% higher than 285K

285K should outperform a 14900K on single core by <12%.

 

[Tarkin77]

Reading through the post, it looks like MebiuW actually said the following:

14900K should clock ~12% higher than 285K

285K should outperform a 14900K on single core by <12%.

https://www.igorslab.de/en/intels-i...n-for-raptor-lake-s-refresh-and-arrow-lake-s/ like igor said almost a year ago

 

[Ghostsonplanets]

Exist also alluded that Igor projections track with what the core accomplished. So this makes sense.

20A is not clocking as high as the 14900ks.

*N3B. 20A is only for 6+8 DT die that will be manufactured on both 20A or N3B.

 

[Hulk]

If we are talking about 5.5GHz all-core max frequency for Arrow Lake and assuming a good increase in transistor density and efficiency over Intel 7, then I am impressed with Intel's first crack at 20A.

Couple of things to keep in mind. All of the ST frequencies advertised for Raptor Lake are pretty much moot outside of benchmarking. First, rarely will an application use only 1 or 2 cores and allow for these frequencies to occur outside of benchmarking. Second, for anything but a golden sample 6GHz requires 1.5V of CPU cooking high energy electrons that will degrade the CPU.

The only meaningful comparison of clock speeds among CPU's is all-core speeds. As I wrote above if ARL can to 5.5GHz all-core vs 5.8GHz for RPL then that is a win for Intel. If 5.5GHz is ST boost and nC is more like 5.2GHz then I'm not as impressed with 20A on the first effort. It would basically be ADL as noted in the article.

I don't think ARL is going to show any MT improvement (in benchmarks) over RPL. The IPC increase in Lion Cove and Skymont I predict will be about 20%, which will result in a "side grade" from RPL to ARL in benchmarking MT performance. But, in real world applications, in which only 8 or 10 cores are utilized we will see bigger gains for ARL due to the compute increase of Lion Cove. The lack of HT will compromise MT performance in perfectly multithreaded applications.

Intel will most likely promote ARL based on real world applications that show nice gains over RPL and avoid Cinebench MT and such. At least until the 8+32 refresh comes out and resolves the lack of HT issue.

 

[Hulk]

Also I like that Intel started their numbering scheme over with the advent of tile based CPU's. MTL are "1" series, ARL "2" series, etc... We don't need to really care about the Ultra as we know the gen/architecture from that first digit.

That as a good move on Intel's part.

 

[Hitman928]

Also I like that Intel started their numbering scheme over with the advent of tile based CPU's. MTL are "1" series, ARL "2" series, etc... We don't need to really care about the Ultra as we know the gen/architecture from that first digit.

That as a good move on Intel's part.

RPL-R are 1 series as well though. . .

 

[mikk]

If we are talking about 5.5GHz all-core max frequency for Arrow Lake and assuming a good increase in transistor density and efficiency over Intel 7, then I am impressed with Intel's first crack at 20A.

It's a N3B node for the highest clocked parts. 20A 6+8 will be clocked lower.

I don't think ARL is going to show any MT improvement (in benchmarks) over RPL. The IPC increase in Lion Cove and Skymont I predict will be about 20%, which will result in a "side grade" from RPL to ARL in benchmarking MT performance. But, in real world applications, in which only 8 or 10 cores are utilized we will see bigger gains for ARL due to the compute increase of Lion Cove. The lack of HT will compromise MT performance in perfectly multithreaded applications.

Intel will most likely promote ARL based on real world applications that show nice gains over RPL and avoid Cinebench MT and such. At least until the 8+32 refresh comes out and resolves the lack of HT issue.

The early performance projection slide from igorslab doesn't seem that wrong what Exist and some others implied, in this case MT gains should be higher over Raptor Lake refresh. Geekbench ST 9-13% and MT 16-20% in this.

 

[AMDK11]

This prediction says absolutely nothing about Arrowlake's clock speed and is only about projected performance at 250W.

 

[mikk]

This prediction says absolutely nothing about Arrowlake's clock speed and is only about projected performance at 250W.

I refer to the performance prejection from some people, they say it's roughly what we get. And the new leak says 5.5 Ghz with maybe less than 12% ST improvement which is in the same ballpark as the projected 9-13% from the early Intel projection. Maybe they projected with 5.5 Ghz from the beginning.

 

[AMDK11]

This is still just speculation. We have absolutely no hard data on the clock speed this prediction refers to.

It looks as if Intel just wanted to show the increase in performance with the same energy consumption.

 

[Ghostsonplanets]

I'm not sure where to share this, but ooof

INVESTOR ALERT: Law Offices of Howard G. Smith Announces the Filing of a Securities Class Action on Behalf of Intel Corporation (INTC) Investors

Law Offices of Howard G. Smith announces that a class action lawsuit has been filed on behalf of investors who purchased Intel Corporation (“Intel” or

www.businesswire.com

Class Action against Intel due to misleading Intel Foundry reports and claims.

 

[Joe NYC]

I'm not sure where to share this, but ooof

INVESTOR ALERT: Law Offices of Howard G. Smith Announces the Filing of a Securities Class Action on Behalf of Intel Corporation (INTC) Investors

Law Offices of Howard G. Smith announces that a class action lawsuit has been filed on behalf of investors who purchased Intel Corporation (“Intel” or

www.businesswire.com

Class Action against Intel due to misleading Intel Foundry reports and claims.

I don't like class action lawsuits in general, but there are some good points in this one.

There is a line between cheerleading and misleading investors, and Gelsinger has been at or beyond that line.

 

[Goop_reformed]

I don't like class action lawsuits in general, but there are some good points in this one.

There is a line between cheerleading and misleading investors, and Gelsinger has been at or beyond that line.

But but what about the push ups?

 

[Hulk]

It's a N3B node for the highest clocked parts. 20A 6+8 will be clocked lower.




The early performance projection slide from igorslab doesn't seem that wrong what Exist and some others implied, in this case MT gains should be higher over Raptor Lake refresh. Geekbench ST 9-13% and MT 16-20% in this.

It will be a magic trick if they can do 16% better MT for 8+16 RPL to 8+16 ARL with no HT in ARL.

 

[Kepler_L2]

It will be a magic trick if they can do 16% better MT for 8+16 RPL to 8+16 ARL with no HT in ARL.

HT doesn't boost performance much on RPL and Skymont is a big jump.

 

[DrMrLordX]

I don't like class action lawsuits in general, but there are some good points in this one.

There is a line between cheerleading and misleading investors, and Gelsinger has been at or beyond that line.

The bigger issue is that Intel is pouring a ton of money into their fabs and foundry service model without having major revenue sources from the design side of the company to back it up. They're bleeding money on poor-to-non-existent margins, especially in DCG. The only thing this class action suit is really saying is "we see all this money being spent on fabs with nothing short term to show for it, and we would rather that money be spent on buybacks or dividends".

All that aside, I think maybe this topic is better situated in the cutting edge fab thread since it's an IFS issue more than a Meteor/Arrow Lake topic.

 

[Hulk]

HT doesn't boost performance much on RPL and Skymont is a big jump.

HT performance boost on MT apps/benches like Blender/Cinebench/Handbrake (some settings) is significant. Like 25%. Granted those are a niche software as I noted previously but the "old" scores from RPL are still there for Intel to deal with.

Let me be specific:
Doesn't have to be a magic trick for you if Intel ARL 8+16 beats RPL 8+16 in Cinebench MT, but it will be for me! The metric is for ARL to achieve 40,000 CB R23 MT points. If that happens I will be impressed considering the clock speed regression, lack of SMT, and tile latency.

 

[Saylick]

The bigger issue is that Intel is pouring a ton of money into their fabs and foundry service model without having major revenue sources from the design side of the company to back it up. They're bleeding money on poor-to-non-existent margins, especially in DCG. The only thing this class action suit is really saying is "we see all this money being spent on fabs with nothing short term to show for it, and we would rather that money be spent on buybacks or dividends".

All that aside, I think maybe this topic is better situated in the cutting edge fab thread since it's an IFS issue more than a Meteor/Arrow Lake topic.

Eh, as long as Uncle Sam is paying them gubmint subsidies, Intel will have the liquidity to keep spending on their Foundry.

HT performance boost on MT apps/benches like Blender/Cinebench/Handbrake (some settings) is significant. Like 25%. Granted those are a niche software as I noted previously but the "old" scores from RPL are still there for Intel to deal with.

Let me be specific:
Doesn't have to be a magic trick for you if Intel ARL 8+16 beats RPL 8+16 in Cinebench MT, but it will be for me! The metric is for ARL to achieve 40,000 CB R23 MT points. If that happens I will be impressed considering the clock speed regression, lack of SMT, and tile latency.

It would be a surprise to me if ARL can’t achieve 40,000 MT score in CB23. Skymont cores are supposed to be significantly improved, which ought to make up for the lack of SMT in the P cores. I also don’t expect Skymont to have a clock regression. The P cores themselves are lackluster, but even they are supposed to have a low-teens improvement. If we consider that RPL already achieves scores in the upper 30,000 range, a 10% improvement in ARL should be the minimum gain imo. Lastly, CB is not sensitive to latency if I remember correctly, so a tiled arch shouldn’t degrade the MT score.

 

[Philste]

considering the clock speed regression

What clock speed regression? We are talking about MT here. 13900K runs at 4.9-5.2GHz in MT. With all the stuff currently happening we should look at the lower range, so 4.9GHz for baseline 253W profile. ARL can hit those clocks in MT as well.

 

[Kepler_L2]

HT performance boost on MT apps/benches like Blender/Cinebench/Handbrake (some settings) is significant. Like 25%. Granted those are a niche software as I noted previously but the "old" scores from RPL are still there for Intel to deal with.

Let me be specific:
Doesn't have to be a magic trick for you if Intel ARL 8+16 beats RPL 8+16 in Cinebench MT, but it will be for me! The metric is for ARL to achieve 40,000 CB R23 MT points. If that happens I will be impressed considering the clock speed regression, lack of SMT, and tile latency.

It's definitely not 25% boost on Intel CPUs. More like 10-15% at best.

 

[uzzi38]

What clock speed regression? We are talking about MT here. 13900K runs at 4.9-5.2GHz in MT. With all the stuff currently happening we should look at the lower range, so 4.9GHz for baseline 253W profile. ARL can hit those clocks in MT as well.

I don't think anything is known about ARL's MT clocks yet. Considering these days Intel achieves as high ST clocks as it does thanks to TB3.0 and TVB, I wouldn't expect ARL to achieve the same ST clocks as MT clocks unless they cut these features entirely.

Even as a best case scenario I'd expect at minimum a ~200MHz clock deficit vs ST boost. By comparison, the 14900KS sees a ~400MHz boost between regular ST boost and TVB boost on the preferred core. MT boost will inevitably see a larger difference to ST clocks again.

 

[Hulk]

What clock speed regression? We are talking about MT here. 13900K runs at 4.9-5.2GHz in MT. With all the stuff currently happening we should look at the lower range, so 4.9GHz for baseline 253W profile. ARL can hit those clocks in MT as well.

Incorrect. 13900K runs 5.5GHz stock in MT.

Intel Core i9-13900K Review - Power-Hungry Beast

With the Core i9-13900K, Intel delivers impressive performance. Our in-depth review confirms: Raptor Lake is the world's fastest CPU for gaming. Even in applications the processor is able to match AMD's Zen 4 Ryzen 9 7950X flagship. If only power consumption wasn't so high...

www.techpowerup.com

Furthermore it is customary to compare a new generation to the last generation, not two generations ago. 14900K will boost to 5.7GHz in stock form. As I wrote, ARL has a lot of ground to make up.

Intel Core i9-14900K Review - Reaching for the Performance Crown

Core i9-14900K is Intel's new flagship with clock speeds of up to 6 GHz. It's actually clocked even higher than the 13900KS, thanks to an extra 100 MHz when more than two cores are active. Our review confirms that Raptor Lake Refresh is amazing for both applications and gaming, if you can live...

www.techpowerup.com

14900KS will actually boost to 5.9GHz all-core but we'll keep the ARL comparison to 14900K.

Intel Core i9-14900KS Review - The Last of its Kind

The Intel Core i9-14900KS reaches impressive clock frequencies up to 6.2 GHz, which yields remarkable performance in applications and gaming. However, this comes at the cost of serious power consumption, especially if you go beyond stock, where we saw over 500 W during testing.

www.techpowerup.com

 

[Hulk]

It's definitely not 25% boost on Intel CPUs. More like 10-15% at best.

Incorrect. RPL will do 30% or more MT increase with HT in Cinebench MT.

Raptor Lake - 10/20/2022
13900K						40760
Raptor Cove P Cores	8	5.8	2256	389	5.5	22528	2816	512	31.6%
Gracemont E Cores	16	4.3	1140	265	4.3	18232	1140	265
13600K			1940	(Actual)		23365	(Actual)
Raptor Cove P Cores	6	5.1	1943	381	5.1	15300	2550	500	31.2%
Gracemont E Cores	8	3.9	1010	259	3.9	8081	1010	259
						23381

 

[Hulk]

Double post error.