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

[Tigerick]

Wildcat Lake (WCL) Preliminary Specs

Intel Wildcat Lake (WCL) is upcoming mobile SoC replacing ADL-N. 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 Q2/Computex 2026. In case people don't remember AlderLake-N, I have created a table below to compare the detail specs of ADL-N and WCL. Just for fun, I am throwing LNL and upcoming Mediatek D9500 SoC.

		Intel Alder Lake - N	Intel Wildcat Lake	Intel Lunar Lake		Mediatek D9500
Launch Date		Q1-2023	Q2-2026 ?	Q3-2024		Q3-2025
Model		Intel N300	?	Core Ultra 7 268V		Dimensity 9500 5G
Dies		2	2	2		1
Node		Intel 7 + ?	Intel 18-A + TSMC N6	TSMC N3B + N6		TSMC N3P
CPU		8 E-cores	2 P-core + 4 LP E-cores	4 P-core + 4 LP E-cores		C1 1+3+4
Threads		8	6	8		8
Max Clock		3.8 GHz	?	5 GHz
L3 Cache		6 MB	?	12 MB
TDP		7 W	Fanless ?	17 W		Fanless
Memory		64-bit LPDDR5-4800	64-bit LPDDR5-6800 ?	128-bit LPDDR5X-8533		64-bit LPDDR5X-10667
Size		16 GB	?	32 GB		24 GB ?
Bandwidth			~ 55 GB/s	136 GB/s		85.6 GB/s
GPU		UHD Graphics		Arc 140V		G1 Ultra
EU / Xe		32 EU	2 Xe	8 Xe		12
Max Clock		1.25 GHz		2 GHz
NPU		NA	18 TOPS	48 TOPS		100 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.

 

[lightisgood]

Next-Gen AMD 700-Series “AM5” & Intel 800-Series “LGA 1851” Socket Motherboards Aiming Q3 2024 Launch

Next-Gen AMD 700-Series "AM5" & Intel 800-Series "LGA 1851" Socket Motherboards Aiming Q3 2024 Launch

AMD & Intel motherboard partners are aiming for a Q3 2024 launch for the next-gen 700-series "AM5" & 800-series "LGA 1851" platforms.

wccftech.com

So, Arrow Lake-S should launch in 3Q24.

 

[Hulk]

Actually, MTL's NOC fabric is extremely power efficient and far more power efficient than AMD's Infinity fabric. The single culprit for higher power draw in MTL is the CPU tile (RWC) itself. Maybe requires a pcode update or is just plain bad considering it is an ageing design. Either way, it's time for RWC to go away.

Before we throw RWC into the trash can let's see some head-to-head power results of RWC on Intel 4 vs. RPC on Intel 7.

 

[igor_kavinski]

Before we throw RWC into the trash can

I'm just trying to understand why people hate it so much. I mean, Hello. No one threw the Skylake derivatives into the trash. In fact, people happily paid money for 10900K CPUs. Everyone hating RWC should just chill. In monolithic form, I'm sure it would've been nice. It just can't put its best foot forward due to Intel desperately trying to make up for all its missteps over the past several years by putting it in a disaggregated tile arch.

 

[SiliconFly]

This is based on what?

Based on the NoC itself. It was acquired by Keller himself I think. One of it's primary goals of the NoC is to reduce SoC power. It's more power efficient than the bus it replaces as the links are somehow packetized like in a LAN. The tile-to-tile NoC links are handled by Foveros which has one of the Industries highest power efficiency. The NoC by itself cannot be the source of the high power draw we see in some of the MTL power usage tests.

 

[SiliconFly]

I'm just trying to understand why people hate it so much. I mean, Hello. No one threw the Skylake derivatives into the trash. In fact, people happily paid money for 10900K CPUs. Everyone hating RWC should just chill. In monolithic form, I'm sure it would've been nice. It just can't put its best foot forward due to Intel desperately trying to make up for all its missteps over the past several years by putting it in a disaggregated tile arch.

I remember seeing a slide posted in this forum by one of the members a while back which clearly showed that RWC is actually more a derivative of GLC rather than RPC. Even including all the power and performance optimizations in RWC, it's still putting a lipstick on a hog, because GLC & RPC are power hogs. Without a doubt, RWC is the weakest link in the (MTL) chain imho.

 

[igor_kavinski]

Without a doubt, RWC is the weakest link in the (MTL) chain imho.

Since Intel announced before Alder Lake that they made their architectures process node agnostic, I wonder why they didn't just bring LNC forward in Meteor Lake? Being lazy and overconfident as usual, I suppose?

 

[SiliconFly]

Since Intel announced before Alder Lake that they made their architectures process node agnostic, I wonder why they didn't just bring LNC forward in Meteor Lake? Being lazy and overconfident as usual, I suppose?

A new architecture like LNC is a monumental task that takes a lot of time I think. RWC development started way too early. If I remember right, MTL CPU tile taped-in long ago. And ARL CPU tile taped-in only very recently I think. Thats a long gap. LNC never stood a chance (with MTL) as Intel was playing it very safe. It's gonna be very interesting when LNC debuts with ARL though.

 

[coercitiv]

I remember seeing a slide posted in this forum by one of the members a while back which clearly showed that RWC is actually more a derivative of GLC rather than RPC. Even including all the power and performance optimizations in RWC, it's still putting a lipstick on a hog, because GLC & RPC are power hogs. Without a doubt, RWC is the weakest link in the (MTL) chain imho.

So, just for the record, your though Intel was going to dominate mobile with lipstick on a power hog?

 

[FlameTail]

Damn. All these abbreviations are getting under my skin.

MTL, ADL, RKL, GLC, RPC, LNL, LNC etc... etc...

 

[SiliconFly]

So, just for the record, your though Intel was going to dominate mobile with lipstick on a power hog?

Not really. Like I keep saying often, the strength of MTL lies in it's LP E cores. The CPU tile should fire up only when needed and shouldn't be much of a problem in everyday workloads. At least, thats the whole idea.

It'll be interesting to see the results when more OEMs release their MTL laptops soon.

 

[SiliconFly]

Damn. All these abbreviations are getting under my skin.

MTL, ADL, RKL, GLC, RPC, LNL, LNC etc... etc...

You missed RPL & RWC.

 

[igor_kavinski]

You missed RPL & RWC.

And PTL/PNC.

 

[FlameTail]

And PTL/PNC.

And the R's

SFR, GNR, EMR, FSR, DMR

 

[coercitiv]

Not really. Like I keep saying often, the strength of MTL lies in it's LP E cores.

Type slower please, you're gonna' wake up the P core...

 

[Tigerick]

Type slower please, you're gonna' wake up the P core...

Make sure he close other tabs, you know with Youtube video running background might wake up LP cores as well

 

[SiliconFly]

Of course the CPU tile will turn on and turn off constantly and I think it'll look more like this in real world scenario. Just sayin...



Overall, the CPU tile power usage shouldn't be a big deal during average everyday use.

 

[PJVol]

Yeah, performace and power are two independent things in itself now

 

[igor_kavinski]

I tested Meteor Lake CPUs on Intel's optimized AI tools: AMD's chips often beat them

The new chips' NPU helps a lot, but AMD's AI accelerated Ryzen CPUs won on some tests.

www.tomshardware.com

MTL NPU loses to Ryzen in a couple tests. But it wins by a good margin too. The real AI match will begin on release of Windows 12.

 

[FlameTail]

I tested Meteor Lake CPUs on Intel's optimized AI tools: AMD's chips often beat them

The new chips' NPU helps a lot, but AMD's AI accelerated Ryzen CPUs won on some tests.

www.tomshardware.com

MTL NPU loses to Ryzen in a couple tests. But it wins by a good margin too. The real AI match will begin on release of Windows 12.

I guess Snapdragon X Elite will be crushing those tests, thanks to the whopping 45 TOPS of NPU and combined 75 TOPS for the SoC.

Atleast until Strix Point and Arrow Lake arrive I guess.

 

[Khato]

Well, as all the NPU benchmarks thus far have shown... Doesn't matter how many dedicated AI TOPS you put in silicon if you don't have meaningful software support.

 

[DrMrLordX]

Next-Gen AMD 700-Series “AM5” & Intel 800-Series “LGA 1851” Socket Motherboards Aiming Q3 2024 Launch

So, Arrow Lake-S should launch in 3Q24.

Intel could put Raptor Lake Refresh on that socket as well. There's no indicator that Arrow Lake will launch before October of next year.

 

[SiliconFly]

Looks like the new pcode update brings in huge power efficiency uplift to MTL...

 

[igor_kavinski]

Intel could put Raptor Lake Refresh on that socket as well.

They shouldn't be allowed to do that unless RPL-R is an Intel 4 refresh. People, please send letters to your senators if Intel tries to sell old CPUs on that new socket.

 

[SiliconFly]

They shouldn't be allowed to do that unless RPL-R is an Intel 4 refresh. People, please send letters to your senators if Intel tries to sell old CPUs on that new socket.

Letters won't cut it. Storm their HQ.

 

[TESKATLIPOKA]

Looks like the new pcode update brings in huge power efficiency uplift to MTL...

Nothing new there, It was already mentioned here.
And you even call It as a huge efficiency uplift.
It's not that big, only 7% higher sustained score in CB R23. And you call It as huge.