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.)

 

[Det0x]

Not sure if this already have been posted, but this is not looking good..
Hopefully it will get alot better when we get closer to launch

 

[Hitman928]

Not sure if this already have been posted, but this is not looking good..
Hopefully it will get alot better when we get closer to launch
View attachment 107659

Doesn't seem to be published anymore (at least that I can find) but there is a 265KF that scores slightly higher. There are also 285Ks that score in the 3420 - 3450 range, which puts it basically in line with the 9950x. It seems the 285K has a noticeable advantage in pts/GHz over the 265K in Geekbench which is different than say the 9700x to 9950x, though maybe memory speeds come into play as well.

 

[OneEng2]

Lunar Lake is on N3B. Arrow Lake is likely on a higher performance variant of N3. Lunar Lake should make much more money for Intel than Arrow Lake, being that its going into premium laptops that command higher margins and is a lot less silicon area than Arrow Lake.

Arrow Lake will likely be expensive, so I have my doubts AMD will lower Zen 5 desktop prices due to Arrow Lake. Zen 5 desktop is AMDs second least profitable business, only consumer /desktop GPU is less profitable. They would rather sell the silicon in EPYC, Instinct/MI, or laptop at higher prices than sell desktop at lower prices.

Undoubtably true. High margin chips are definitely where AMD would choose to sell; however, the ZEN 5 Turin will be produced on the N3E (I believe). While the desktop processor shares the same core architecture, it doesn't share fab lines therefore they can't really utilize the Turin dies to sell in the desktop/laptop or vise versa. Note, AMD does have an interesting follow-on product with an N3E version of their ZEN 5 desktop processor.

AFAIK Intel still doesn't have anything in the server space that will compete with Turin, so I doubt AMD will be hurting in the profit department overall. To that end, I agree with you that AMD is unlikely to lower their desktop and laptop processor prices even though their cost is likely much lower than Intel's to produce. I am not sure that AMD would want to give up the desktop and or laptop market share to Intel, but that would be a strategic decision on their part if the price and performance pressure from Intel forces them into lowering price to keep market share.

I do agree that Arrow Lake will likely have to be more expensive than Zen 5 for desktop and laptop parts due to the more expensive process node being used.

 

[Henry swagger]

Not sure if this already have been posted, but this is not looking good..
Hopefully it will get alot better when we get closer to launch
View attachment 107659

You concern trolling stop it

 

[Hitman928]

Undoubtably true. High margin chips are definitely where AMD would choose to sell; however, the ZEN 5 Turin will be produced on the N3E (I believe). While the desktop processor shares the same core architecture, it doesn't share fab lines therefore they can't really utilize the Turin dies to sell in the desktop/laptop or vise versa. Note, AMD does have an interesting follow-on product with an N3E version of their ZEN 5 desktop processor.

AFAIK Intel still doesn't have anything in the server space that will compete with Turin, so I doubt AMD will be hurting in the profit department overall. To that end, I agree with you that AMD is unlikely to lower their desktop and laptop processor prices even though their cost is likely much lower than Intel's to produce. I am not sure that AMD would want to give up the desktop and or laptop market share to Intel, but that would be a strategic decision on their part if the price and performance pressure from Intel forces them into lowering price to keep market share.

I do agree that Arrow Lake will likely have to be more expensive than Zen 5 for desktop and laptop parts due to the more expensive process node being used.

Only Turin-D uses N3E, regular Turin uses N4P like desktop Zen 5. I personally haven’t heard about a desktop N3E version of Zen 5.

 

[Nothingness]

Doesn't seem to be published anymore (at least that I can find) but there is a 265KF that scores slightly higher. There are also 285Ks that score in the 3420 - 3450 range, which puts it basically in line with the 9950x. It seems the 285K has a noticeable advantage in pts/GHz over the 265K in Geekbench which is different than say the 9700x to 9950x, though maybe memory speeds come into play as well.

The result might be here. But it looks like the search engine DB is updated once a day (or even less often), so too recent results can't be found unless you have a direct link.

 

[OneEng2]

Everything is N3B. In the final stage before being cancelled, the 20A tile was a small volume 6+8. ARL + LNL generation was always planned to be manufactured on TSMC N3 process.

Thanks.

So Intel is paying for the most expensive TSMC process while AMD is paying for N5 which is arguably 2 steps less good than N3B (in density and performance).

 

[SiliconFly]

Thanks.

So Intel is paying for the most expensive TSMC process while AMD is paying for N5 which is arguably 2 steps less good than N3B (in density and performance).

Nope. Zen 5 uses the more advanced N4P process node, which is better than N4 itself I think. Very close to N3B.

 

[Hitman928]

Thanks.

So Intel is paying for the most expensive TSMC process while AMD is paying for N5 which is arguably 2 steps less good than N3B (in density and performance).

AMD is using N4P for consumer products and Turin (regular), which is really close in performance and efficiency to N3B (listed as just N3 in the chart below). There is a significant logic density advantage for N3B, but across the whole chip and with the way Intel designs seem to go, the density advantage doesn't translate all that well to an area advantage for Intel.

 

[511]

Thanks.

So Intel is paying for the most expensive TSMC process while AMD is paying for N5 which is arguably 2 steps less good than N3B (in density and performance).

Yes and No it depends on the contract negotiatied and the volume Intel's volume is way larger and second only to Apple according to estimates Chiplets are cost effective as well anyway it wull erode their CCG Margin

 

[9949asd]

Not sure if this already have been posted, but this is not looking good..
Hopefully it will get alot better when we get closer to launch
View attachment 107659

We have 285k 5.7 st 3450, so 265k 5.5 should be 3320.

 

[OneEng2]

Only Turin-D uses N3E, regular Turin uses N4P like desktop Zen 5. I personally haven’t heard about a desktop N3E version of Zen 5.

Thanks for the clarification.

 

[OneEng2]

Yes and No it depends on the contract negotiatied and the volume Intel's volume is way larger and second only to Apple according to estimates Chiplets are cost effective as well anyway it wull erode their CCG Margin

Interesting point. So they will be bidding against Apple for process time.

 

[511]

Interesting point. So they will be bidding against Apple for process time.

Nope no one is bidding against Apple with TSMC they both co develop the nodes but Intel's Volume of Silicon is Nuts

 

[OneEng2]

AMD is using N4P for consumer products and Turin (regular), which is really close in performance and efficiency to N3B (listed as just N3 in the chart below). There is a significant logic density advantage for N3B, but across the whole chip and with the way Intel designs seem to go, the density advantage doesn't translate all that well to an area advantage for Intel.

View attachment 107663

Thanks.

Really close in power, but considerably different in transistor density (ie cost to produce a chip). The performance figures seem strange to me though.

How can N4P be 22% better in power or 11% better in performance while N3 is 25-30% lower in power or 10-15% better in performance? It was my understanding that when they quote power, they assume the same performance and when they quote performance, them assume at the same power level as the previous node?

It still looks like N3 is a considerably better process node than N4P .... just more expensive.

I am also confused why Intel is not using N3E which is supposed to be superior in performance and power than N3(B), and less expensive but at the expense of die size .... unless the die size cost outweighs the less expensive price per area of the N3E process over N3B.

Also, it seems a dubious strategy for Intel to hook its wagon to the N3B process and fight for capacity with Apple on a dead process node as it seems no one else is likely to sign up for N3B moving forward as other better 3nm process nodes are now available that are less expensive and better in most ways than N3B.

 

[OneEng2]

Nope no one is bidding against Apple with TSMC they both co develop the nodes but Intel's Volume of Silicon is Nuts

More nuts than Apple?

 

[OneEng2]

We have 285k 5.7 st 3450, so 265k 5.5 should be 3320.

Do you happen to know what 9550 scores are in comparison?

 

[Josh128]

Undoubtably true. High margin chips are definitely where AMD would choose to sell; however, the ZEN 5 Turin will be produced on the N3E (I believe). While the desktop processor shares the same core architecture, it doesn't share fab lines therefore they can't really utilize the Turin dies to sell in the desktop/laptop or vise versa. Note, AMD does have an interesting follow-on product with an N3E version of their ZEN 5 desktop processor.

AFAIK Intel still doesn't have anything in the server space that will compete with Turin, so I doubt AMD will be hurting in the profit department overall. To that end, I agree with you that AMD is unlikely to lower their desktop and laptop processor prices even though their cost is likely much lower than Intel's to produce. I am not sure that AMD would want to give up the desktop and or laptop market share to Intel, but that would be a strategic decision on their part if the price and performance pressure from Intel forces them into lowering price to keep market share.

I do agree that Arrow Lake will likely have to be more expensive than Zen 5 for desktop and laptop parts due to the more expensive process node being used.

Only the Zen 5C variants are thought to use 3nm, the non C EPYCs will use 4nm just like desktop and top out at 96 to 128 core.

Also, its never been confirmed that AMD uses N4P or N4X for Zen 5, afaik.

 

[Hitman928]

Thanks.

Really close in power, but considerably different in transistor density (ie cost to produce a chip). The performance figures seem strange to me though.

That is logic density only (does not include cache or analog/PHY density) and the improvement is significantly less when looking at overall chip density. Additionally, Intel pushes for high frequency designs which will also nullify some of that density advantage when looking at competing designs. That's why I said Intel's chips don't really show a chip area advantage compared to the competition, despite being on N3B versus N4P. The N3B node being significantly more expensive is obviously also not great.

How can N4P be 22% better in power or 11% better in performance while N3 is 25-30% lower in power or 10-15% better in performance? It was my understanding that when they quote power, they assume the same performance and when they quote performance, them assume at the same power level as the previous node?

Yes, it is better power at the same performance or better performance at the same power, you don't get both simultaneously.

It still looks like N3 is a considerably better process node than N4P .... just more expensive.

It's only considerably better in logic density, but I covered that in the first part of the reply.

I am also confused why Intel is not using N3E which is supposed to be superior in performance and power than N3(B), and less expensive but at the expense of die size .... unless the die size cost outweighs the less expensive price per area of the N3E process over N3B.

Also, it seems a dubious strategy for Intel to hook its wagon to the N3B process and fight for capacity with Apple on a dead process node as it seems no one else is likely to sign up for N3B moving forward as other better 3nm process nodes are now available that are less expensive and better in most ways than N3B.

Story is that Intel had to commit to buying significant wafers from TSMC far in advance to get access to a leading edge node because TSMC didn't want to be a "crutch" for Intel and get the rug pulled out if Intel decided to stick with their own fabs. So, Intel had to make a choice to either commit to buying significant amounts of N3B (just N3 at the time) or risk relying solely on their own fabs and potentially having another 10 nm fiasco on their hands. They decided to commit to buying N3B wafers.

Do you happen to know what 9550 scores are in comparison?

Roughly the same. Depends on the memory config, but with decent memory (i.e., DDR5-6000 or higher and decent timings) it scores basically the same.

 

[511]

More nuts than Apple?

If only Apple than probably Yes.i am not sure though it may be close

 

[9949asd]

Do you happen to know what 9550 scores are in comparison?

Almost the same, but you need to know zen5’s avx512 will bump up the score about 100-150.

 

[Hitman928]

Almost the same, but you need to know zen5’s avx512 will bump up the score about 100-150.

No, AVX512 has basically no effect on Zen GB6 scores.

Page 609 - Discussion - Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)

Page 609 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

 

[Det0x]

Do you happen to know what 9550 scores are in comparison?

This one is mine, but it is overclocked so its prob not a valid comparison (but if that 285k was running cudimms @ 9000MT/s i would say it is)

*edit*

Here are one more 285K score
I really hope its just the MSI motherboard that's underperforming here

https://videocardz.com/newz/intel-core-ultra-9-285k-and-ultra-7-265kf-arrow-lake-cpus-geekbench-scores-featuring-msi-z890-pro-motherboard-have-been-leaked

 

[jdubs03]

This one is mine, but it is overclocked so its prob not a valid comparison (but if that 285k was running cudimms @ 9000MT/s i would say it is)

*edit*

Here are one more 285K score
I really hope its just the MSI motherboard that's underperforming here
View attachment 107668

https://videocardz.com/newz/intel-core-ultra-9-285k-and-ultra-7-265kf-arrow-lake-cpus-geekbench-scores-featuring-msi-z890-pro-motherboard-have-been-leaked

That would be very underwhelming. But it also is much lower than 3 previous leaks which all perform much better, and to me would be more representative of performance.

 

[alcoholbob]

This looks like maybe 10% ST and 5% better MT than a 14900K?