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.)

 

[A///]

Buy two 7950X3D's, the other one with the cheapest mobo you can find and you could do the same amount of encoding work (maybe more) than those crappy hot Intels. I have a 12700K and I wouldn't encode on it for too long. It just gets too hot for comfort. Not to mention inefficient. Sell or donate your Intel PCs.

Most likely the viable solution. The 13900K are significantly better but as I said they run hotter and restraining them to 253 watts is level with AMD, the cache won't do much for software encodes but there's small benefits to the amd platform. I can get away by running larger ram kits at reduced speed. I don't know if I'll donate them but I'm sure I can find a use for them eventually if not sell them to someone for cheap. I got them dirt cheap through combo deals online. I'm expecting AMD to move head first into ai/neural and inferencing with zen 5 and beyond as Intel plans to do the same.

 

[igor_kavinski]

the cache won't do much for software encodes

Sure about that?

AMD Ryzen 9 7950X3D Review - KitGuru

As KitGuru prepared for CES 2023 we knew full well that AMD would announce a version of Ryzen 7000 w

www.kitguru.net

7950X3D close to 7950X at just 151W. The cache is doing SOMETHING.

 

[A///]

Sure about that?

AMD Ryzen 9 7950X3D Review - KitGuru

As KitGuru prepared for CES 2023 we knew full well that AMD would announce a version of Ryzen 7000 w

www.kitguru.net

7950X3D close to 7950X at just 151W. The cache is doing SOMETHING.

????

save 52 watts and lose roughly 4 frames processing power? That's it? I'd be impressed if the 3d can shave off minutes of encoding time per file. Or in my case, a few hours off each batch.

 

[igor_kavinski]

????

save 52 watts and lose roughly 4 frames processing power? That's it? I'd be impressed if the 3d can shave off minutes of encoding time per file. Or in my case, a few hours off each batch.

You could tune it to do more.

 

[RTX2080]

ARL and LNL preliminary support, LNL-M/R sooner than LNL-S?

https://twitter.com/x/status/1640577467571380224

 

[Tigerick]

Hmm, LNL supposedly should scale down from 15W TDP to 9-12W in order to support fanless design. Especially with 442 max config, there is no chance to fight with ARL-P with 683 config...

 

[jpiniero]

there is no chance to fight with ARL-P with 683 config...

Intel does that now with the 2+8 die.

 

[Tigerick]

That is U series, RPL-P comes with 4+8 or even 6+8 config which I believe will be standard for upcoming ARL-P, that's why I said LNL-P doesn't make sense, we shall see

Intel does that now with the 2+8 die.

 

[jpiniero]

That is U series, RPL-P comes with 4+8 or even 6+8 config which I believe will be standard for upcoming ARL-P, that's why I said LNL-P doesn't make sense, we shall see

That's because there's no i3 P with Raptor Lake.

 

[hemedans]

That's because there's no i3 P with Raptor Lake.

There is i3 1320PE with 4P+4E

 

[mikk]

ARL and LNL preliminary support, LNL-M/R sooner than LNL-S?

LNL is mobile only. The next after Arrow Lake-S should be Panther Lake.

 

[A///]

LNL is mobile only. The next after Arrow Lake-S should be Panther Lake.

Made with real extract of panther I hope!

 

[Exist50]

The foundation of lowering system power comes with Meteorlake. The issue is that the PCH cannot power down fast enough in realistic scenarios, since even in Icelake it can reach ultra low standby power. What's the point of super low standby if it can only be reached literally after 2 mins of doing absolutely nothing, not even moving a mouse pointer?

This problem has been "known" for a decade now. I think if they did not screw up royally on the 10nm generation, we would have seen it quite a bit before 2020.

They might have a lot to perfect still, and that could be Lunar Lake.

If Intel is going to see a significant increase in battery life, it will come with Lunar Lake or not at all. I doubt Meteor Lake will provide a significant improvement.

 

[A///]

Meteor is supposed to bring great performance according to gelsinger gunslinger. don't recall battery performance being brought up.

 

[nicalandia]

Meteor is supposed to bring great performance according to gelsinger gunslinger. don't recall battery performance being brought up.

The last leak posted by YuuKi_Ans would indicate a 12-15% performance boost vs Raptor Cove, I suspect that is 10% IPC + 5 speed on Lion Cove cores(being conserbative, it could 8 IPC 7% core speed)

 

[Markfw]

The last leak posted by YuuKi_Ans would indicate a 12-15% performance boost vs Raptor Cove, I suspect that is 10% IPC + 5 speed on Lion Cove cores(being conserbative, it could 8 IPC 7% core speed)

View attachment 78776

OK, is it me or is there nothing to touch Genoa (which is already out, I have one building, 9654, 96 core, 12 channel memory). What is Intel thinking ?

 

[nicalandia]

OK, is it me or is there nothing to touch Genoa (which is already out, I have one building, 9654, 96 core, 12 channel memory). What is Intel thinking ?

Intel will not be able to Match Genoa in the next 2 years, Granite Rapids is a 2022 product that will be release 2024. Itel Server CPUs have been delayed(thanks to Sapphire Rapids) for two years, Sapphire Rapids was scheduled to be released right after Rome and before Milan, Emerald Rapids was to be a direct competitor to Milan/Milan-X and Granite Rapids/Sierra Forest a competitor to Genoa/Bergamo. That's just not feasable any more, they just cant expedite Granite Rapids release

 

[Markfw]

Intel will not be able to Match Genoa in the next 2 years, Granite Rapids is a 2022 product that will be release 2024. Itel Server CPUs have been delayed(thanks to Sapphire Rapids) for two years, Sapphire Rapids was scheduled to be released right after Rome and before Milan, Emerald Rapids was to be a direct competitor to Milan/Milan-X and Granite Rapids/Sierra Forest a competitor to Genoa/Bergamo. That's just not feasable any more, they just cant expedite Granite Rapids release

But I mean granite rapids does not seem to compete with Genoa on any level. Or I am misreading this leak ??

 

[Geddagod]

The last leak posted by YuuKi_Ans would indicate a 12-15% performance boost vs Raptor Cove, I suspect that is 10% IPC + 5 speed on Lion Cove cores(being conserbative, it could 8 IPC 7% core speed)

View attachment 78776

Explain that estimation please?

 

[Geddagod]

But I mean granite rapids does not seem to compete with Genoa on any level. Or I am misreading this leak ??

Do you think 2X SPR performance is not enough to beat Genoa?

 

[Markfw]

Do you think 2X SPR performance is not enough to beat Genoa?

I have not seen enough leaks to believe that, no. Show me some please. SPR is so bad from what I have seem that I don't think so. And AMD is not standing still either.

Edit: the only leak I have seen is that 13900k beats the 2495x, and maybe even the 3495x Xeon. This looks pitiful.

 

[A///]

The last leak posted by YuuKi_Ans would indicate a 12-15% performance boost vs Raptor Cove, I suspect that is 10% IPC + 5 speed on Lion Cove cores(being conserbative, it could 8 IPC 7% core speed)

View attachment 78776

Mind telling me where i'm supposed to look on a granite rapids leak sheet? Although this may be the first time someone's relied on aggregate performance as fact... lol

 

[A///]

OK, is it me or is there nothing to touch Genoa (which is already out, I have one building, 9654, 96 core, 12 channel memory). What is Intel thinking ?

They have outstanding contracts so there are customers buying this junk. A lot of this was supposed to come out a while back regardless of whatever bs pushed out that it's actually on time. Intel and their partners likely have a kill fee attached to the contract. Could cost Intel hundreds of millions if they don't deliver.

 

[A///]

Explain that estimation please?

It's aggregate performance. Pure speculation based on frequency to core count among other factors. It's as good as pulling random stats out of your behind. It's used professionally for a really jank estimate of what something may perform like. If the figures were remotely true Intel's desktop would be dead in the water again for years to come given what AMD wants to do.

 

[Geddagod]

I have not seen enough leaks to believe that, no. Show me some please. SPR is so bad from what I have seem that I don't think so. And AMD is not standing still either.

Edit: the only leak I have seen is that 13900k beats the 2495x, and maybe even the 3495x Xeon. This looks pitiful.

Literally on the leak you were quoting, it says Granite Rapids AP, targeting ~2X performance of EMR-64C.
And also I'm pretty sure, at the very least, 2X SPR is enough to beat Genoa.
SPR 56C scores ~65K on CBR23 MT while Genoa scores 115K apparently.
In GB5, 2X the 56C SPR would be 108% of top end Genoa as well.
These are the only MT scores I can find comparing the two. If you can find some that show otherwise, I would be happy to hear them....