Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads

[Tigerick]

Wildcat Lake (WCL) Specs

Intel Wildcat Lake (WCL) is upcoming mobile SoC replacing Raptor Lake-U. WCL consists of 2 tiles: compute tile and PCD tile. It is true single die consists of CPU, GPU and NPU that is fabbed by 18-A process. Last time I checked, PCD tile is fabbed by TSMC N6 process. They are connected through UCIe, not D2D; a first from Intel. Expecting launching in Q1 2026.

		Intel Raptor Lake U	Intel Wildcat Lake 15W?	Intel Lunar Lake		Intel Panther Lake 4+0+4
Launch Date		Q1-2024	Q2-2026	Q3-2024		Q1-2026
Model		Intel 150U	Intel Core 7	Core Ultra 7 268V		Core Ultra 7 365
Dies		2	2	2		3
Node		Intel 7 + ?	Intel 18-A + TSMC N6	TSMC N3B + N6		Intel 18-A + Intel 3 + TSMC N6
CPU		2 P-core + 8 E-cores	2 P-core + 4 LP E-cores	4 P-core + 4 LP E-cores		4 P-core + 4 LP E-cores
Threads		12	6	8		8
Max Clock		5.4 GHz	?	5 GHz		4.8 GHz
L3 Cache		12 MB		12 MB		12 MB
TDP		15 - 55 W	15 W ?	17 - 37 W		25 - 55 W
Memory		128-bit LPDDR5-5200	64-bit LPDDR5	128-bit LPDDR5x-8533		128-bit LPDDR5x-7467
Size		96 GB		32 GB		128 GB
Bandwidth				136 GB/s
GPU		Intel Graphics	Intel Graphics	Arc 140V		Intel Graphics
RT		No	No	YES		YES
EU / Xe		96 EU	2 Xe	8 Xe		4 Xe
Max Clock		1.3 GHz	?	2 GHz		2.5 GHz
NPU		GNA 3.0	18 TOPS	48 TOPS		49 TOPS

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.

 

[Thibsie]

You missed also using Intel fans. Its across all companies.

Or Apple zealots as they were often called in Jobs era.

 

[511]

You missed also using Intel fans. Its across all companies.

there hasn't been one CEO like that since the O.G. Trinity they were humble

 

[DavidC1]

there hasn't been one CEO like that since the O.G. Trinity they were humble

I speculate once a reign of a fantastic CEO is over, all his bad traits essentially get carried over to the future yet the company doesn't abandon it because it was established as a foundation. Such warts and problems with the fantastic CEO works for *him* because it's counteracted by his good qualities and the way he runs the business, but once he's gone, then it's all negative.

The reason I say this is Intel is internally competitive to the point of being detrimental to the operation. The top team is all about squashing everyone that threatens them. It sounds very much like part of Andy Grove's character, a bit of a flaw that got carried into the future. I read an account where Grove wanted to basically stomp competition(AMD) by pricing them out and Moore prevented him from doing so.

 

[511]

I speculate once a reign of a fantastic CEO is over, all his bad traits essentially get carried over to the future yet the company doesn't abandon it because it was established as a foundation. Such warts and problems with the fantastic CEO works for *him* because it's counteracted by his good qualities and the way he runs the business, but once he's gone, then it's all negative.

The reason I say this is Intel is internally competitive to the point of being detrimental to the operation. The top team is all about squashing everyone that threatens them. It sounds very much like part of Andy Grove's character, a bit of a flaw that got carried into the future. I read an account where Grove wanted to basically stomp competition(AMD) by pricing them out and Moore prevented him from doing so.

Yes I am not surprised Huang feared that Intel too brutal and cut throat. Gordon and Noyce were Humble it's sad such great people don't run business anymore 🥹

 

[511]

https://twitter.com/x/status/1878270573870280970

LMAO for all it's fault 14900 is an OC dream

 

[igor_kavinski]

The best CEO is one who has all the potential but never really was a CEO.

Jim Keller

His company seems to be doing fine. But being CEO of Tenstorrent is kinda like being the CEO of a chain of stores in comparison to giants like AMD, Nvidia and former giant Intel. His skills are being under-utilized.

 

[511]

The best CEO is one who has all the potential but never really was a CEO.

Jim Keller

His company seems to be doing fine. But being CEO of Tenstorrent is kinda like being the CEO of a chain of stores in comparison to giants like AMD, Nvidia and former giant Intel. His skills are being under-utilized.

yeah he is a good candidate
btw

Intel® Core™ Processors, FPGAs, GPUs, Networking, Software

Browse Intel product information for Intel® Core™ processors, Intel® Xeon® processors, Intel® Arc™ graphics and more.

www.intel.com

Best ARL CPU?

 

[Thunder 57]

yeah he is a good candidate
btw

Intel® Core™ Processors, FPGAs, GPUs, Networking, Software

Browse Intel product information for Intel® Core™ processors, Intel® Xeon® processors, Intel® Arc™ graphics and more.

www.intel.com

View attachment 114808
Best ARL CPU?

I'm surprised they admit to using TSMC.

 

[igor_kavinski]

I'm surprised they admit to using TSMC.

It's more of a brag I think.

"We using N3B while AMD stuck on N4".

 

[igor_kavinski]

Best ARL CPU?

Cache size is pretty low.

 

[511]

I'm surprised they admit to using TSMC.

Same should be cutdown 6+8 dies lol harvesting dies classic Intel

 

[511]

Cache size is pretty low.

Ahh it is for miniPCs and stuff

 

[Thibsie]

It's more of a brag I think.

"We using N3B while AMD stuck on N4".

I would not brag about that given the circumstances, lol

 

[deasd]

View attachment 114808
Best ARL CPU?

Cache size is pretty low.

It depends on the price.

https://twitter.com/x/status/1878068341983809688

price of U3 205 is not listed but I guess it would be priced nearly ~$200 which is too high.

 

[511]

It depends on the price.

https://twitter.com/x/status/1878068341983809688

price of U3 205 is not listed but I guess it would be priced nearly ~$200 which is too high.

View attachment 114810

if it's $150 its fine than

 

[mikk]

I said on package memory(not enough alone but LNL has PHY optimizations) alone is enough for 0.5-1W reduction which is substantial.

20-30% reduction makes it closer to Meteorlake than Lunarlake, and that's a huge amount. You are going from 15 hours to 10-12 hours.

Did you see the power comparisons from video playback workloads between Lunar Lake and Meteor Lake? It was like 5-6W on LNL versus 10+ watts on Meteor Lake. This is coming from the much stronger LPE core cluster integrated into the compute tile. They can shut down the ringbus cores for many more workloads, this is the real big deal. Darkmont might be a few percent faster than Skymont. In many workloads the LPE cluster should be fast enough. This is not doable on MTL or ARL, even though ARL seems much better under full load thanks to N3B and LNC+Skymont. Under light load however no chance against LNL and for sure PTL.

 

[OneEng2]

Did you see the power comparisons from video playback workloads between Lunar Lake and Meteor Lake? It was like 5-6W on LNL versus 10+ watts on Meteor Lake. This is coming from the much stronger LPE core cluster integrated into the compute tile. They can shut down the ringbus cores for many more workloads, this is the real big deal. Darkmont might be a few percent faster than Skymont. In many workloads the LPE cluster should be fast enough. This is not doable on MTL or ARL, even though ARL seems much better under full load thanks to N3B and LNC+Skymont. Under light load however no chance against LNL and for sure PTL.

It is a nice architectural enhancement IMO.

Using a 3rd tier of processors (LPE) for the lightest of loads for good battery life ..... not a bad idea at all. This (as you point out) allows them to completely shut down the ring bus under these conditions.

 

[DavidC1]

Under light load however no chance against LNL and for sure PTL.

This is in theory, which didn't work well at all for Meteorlake in practice. Intel themselves showed a mere 0.15W savings, which is minutes. I don't know how that video justified itself.

Also 5W and 10W is extremely high for a video playback workload. My Kabylake is under 4.5W while in Youtube playback, so what happened to the 10W system?

There's a huge difference between Lunar and Panther, and that LPE is on another tile. Data needs to be handed off to the LPE before the compute can be turned off, thus there will be a brief period when both tiles are on. If the turn on/off is too slow, you potentially have a scenario where you are better off not having LPE at all.

 

[LightningZ71]

I wonder if the more performant and efficient Arrow lake-U (meteor lake u refresh) has a notably improved power profile than OG Meteor Lake? It seems that it should do better in race to idle and in light load situations. Intel 3 may not be night and day better than 4, but it IS an improvement.

 

[DavidC1]

I wonder if the more performant and efficient Arrow lake-U (meteor lake u refresh) has a notably improved power profile than OG Meteor Lake? It seems that it should do better in race to idle and in light load situations. Intel 3 may not be night and day better than 4, but it IS an improvement.

It can't be worse, can it? It at least clocks significantly higher.

I initially was expecting much more from Foveros regarding latency and power. Maybe some of the disappointment is due to their lack of execution, but products seem to speak for themselves. Lakefield wasn't anything special, neither was Ponte Vecchio, nor Meteorlake.

The fact that there are multiple successors and with big changes to original Foveros suggest there's a way to go before coming even close to fulfilling original promises - disaggregation while minimizing penalty.

 

[DavidC1]

You know their insistence on having just 2 years for platform support is very short sighted.

I think if you look at their dGPU marketshare, it's probably closer to 50/50 split versus 80/20 for their whole client CPU share. With AMD if you have an old enough CPU to seriously bottleneck ARC, you could simply just swap out the CPU, a much better value proposition. AMD CPUs may be "saving" Intel's GPU.

Also, markets with upgradeability shows much more resilience in times of difficulty, because it's much cheaper. That's why notebooks drop far more during times of recession than do desktops.

The whole needing to swap entire platform to upgrade is in line with the mindset of using debt to fuel purchases. When it crashes, it will crash HARD.

 

[511]

I wonder if the more performant and efficient Arrow lake-U (meteor lake u refresh) has a notably improved power profile than OG Meteor Lake? It seems that it should do better in race to idle and in light load situations. Intel 3 may not be night and day better than 4, but it IS an improvement.

Yes I want to see it as well btw I3 has a huge leakage improvement vs Intel 4 so it should be better rest is up to benchmark

 

[adroc_thurston]

https://twitter.com/x/status/1878707839310737542

It got...
...slower? Why. How?

 

[Hitman928]

First Chinese ARL-H review/preview.

https://twitter.com/x/status/1878567395511664845

 

[Hitman928]

More tests. Seems the ARL-H laptop is running at a much higher TDP than the rest in these tests (based on R23 nT score).

https://twitter.com/x/status/1878568819842859222