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.

 

[igor_kavinski]

But its really strange to me that these mobile chips have so high latency compared desktop

Due to soldered LPDDR5 or 5X.

 

[Magio]

It's not an excuse for advertising an unfair comparison, but in which scenarios would you actually be running AI workloads on the CPU, when you have an NPU and GPU available? Seems counter productive both in power consumption and performance.

 

[csbin]

Yeah seems like you are correct 👍
View attachment 106802

But its really strange to me that these mobile chips have so high latency compared desktop, this is one of my old runs on regular Zen5
(on latest tune i haven't made made public yet, i'm sub 74ms)
View attachment 106803

This question still remains, will be interesting when we get closer to launch

AIDA64 and Clam memory are not comparable.

 

[Kepler_L2]

Yeah seems like you are correct 👍
View attachment 106802

But its really strange to me that these mobile chips have so high latency compared desktop, this is one of my old runs on regular Zen5
(on latest tune i haven't made made public yet, i'm sub 74ms)
View attachment 106803

This question still remains, will be interesting when we get closer to launch

MTL memory latency with DDR5 is around 85-90 ns IIRC.

 

[Det0x]

AIDA64 and Clam memory are not comparable.

i'm comparing clam to clam, bottom left of the screenshot i posted

 

[csbin]

i'm comparing clam to clam, bottom left of the screenshot i posted
View attachment 106805

It's safe to say that Intel's lnl latency tests are wrong!

LPDDR5X 8533 should have ~110ns+ latency

Qualcomm’s Oryon Core: A Long Time in the Making

In 2019, a startup called Nuvia came out of stealth mode.

chipsandcheese.com

 

[Hitman928]

i'm comparing clam to clam, bottom left of the screenshot i posted
View attachment 106805

LPDDR has higher latency. LNL I believe uses LPDDRX which may account for some or most of the difference, not sure.

 

[MS_AT]

LPDDR has higher latency. LNL I believe uses LPDDRX which may account for some or most of the difference, not sure.

The packaging should also bring latency benefits I guess. I mean memory is on package, doesn't need to go through motherboard PCB. It's physically much closer to the CPU than otherwise would be possible.

 

[Hitman928]

The packaging should also bring latency benefits I guess. I mean memory is on package, doesn't need to go through motherboard PCB. It's physically much closer to the CPU than otherwise would be possible.

The latency added by the additional distance is insignificant. You may get a small benefit if you don’t have to buffer as much due to the close proximity and reduced parasitics.

 

[9949asd]

The results are pretty much aligned with expectations. 265K should have around the same performance of the 13/14900K.

265k real performance 360aio r23 38k at 180w 60+c。

 

[511]

265k real performance 360aio r23 38k at 180w 60+c。

Where is my screenshot to back it up with HWinfo👀

 

[9949asd]

Where is my screenshot to back it up with HWinfo👀

No screenshot, no need to back it up, soon we will know it.

 

[igor_kavinski]

Where is my screenshot to back it up with HWinfo👀

That result is from a top secret subterranean lab. The message was delivered via note slipped out through a guard and a brave individual risked his life, trekking over miles of rocky terrain littered with hungry hyenas and venomous critters to make the post possible. Go easy on him!

 

[jdubs03]

That result is from a top secret subterranean lab. The message was delivered via note slipped out through a guard and a brave individual risked his life, trekking over miles of rocky terrain littered with hungry hyenas and venomous critters to make the post possible. Go easy on him!

Hyenas on rocky terrain. I MUST BE TRIPPIN.
Everything else checks out that though.

 

[dullard]

But its really strange to me that these mobile chips have so high latency compared to desktop, this is one of my old runs on regular Zen5
(on latest tune i haven't made made public yet, i'm sub 74ms)

LPDDR has higher bandwidth but worse latency than DDR. It is difficult to find CAS latency values for LPDDR5, but if the CL was 39* instead of your CL of 32 (22% more latency), that would explain all the difference. Here is Anandtech's article on it (which was copied from another article since even Anandtech didn't have permission to download the JEDEC timing standards).

AnandTech Forums: Technology, Hardware, Software, and Deals

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

www.anandtech.com

Extended Latencies - LPDDR5X SDRAM devices support extended Read, Write, nWR, ODTLon and ODTLoff Latency Values to account for longer number of cycle it takes to do the data access to memory array.

* There is an AI page that you can find that estimates CL40 for LPDDR5, but I didn't find that credible enough to put too much emphasis on it. I'll just leave this as an IF statement: if the CAS latency is 39.

 

[igor_kavinski]

but if the CL was 39 instead of your CL of 32 (22% more latency), that would explain all the difference.

LPDDR5X-7200 can work at CL28 in CAMM2 form factor: https://www.tomshardware.com/pc-com...tosses-expert-ai-branding-in-for-good-measure

Seems Intel could be keeping costs low (for themselves) with lower quality on-package RAM?

 

[dullard]

LPDDR5X-7200 can work at CL28 in CAMM2 form factor: https://www.tomshardware.com/pc-com...tosses-expert-ai-branding-in-for-good-measure

Seems Intel could be keeping costs low (for themselves) with lower quality on-package RAM?

That would work out to be, what, CL34 at higher frequencies?

I don't really know what the latency is for Lunar Lake. But it is possible that Intel is going with cheaper (and/or more available) higher latency RAM rather than cutting edge RAM. My point is just that people posting above can't claim the latency tests are wrong when we don't even know the latency of the memory being used.

 

[controlflow]

Looks like some machines are popping up for purchase

Buy New ASUS AI Laptop|Discover the best OLED Laptops at Affordable Prices| Now available at ASUS Store

Powered by AI Experience unparalleled performance on Windows 11 Home. Featuring an ultra-fast Intel® Core™ Ultra 7 Series 2 processor with a dedicated AI NPU for 47 TOPs and Intel® Arc™ graphics, the S 14 performs at a higher efficiency and accelerates local AI processes like never before...

shop.asus.com

They can't ship until another 20 days however.

I am curious if these machines with bright high refresh rate OLEDs end up getting poor battery life even with an efficient SOC. Seem like pretty power hungry displays.

 

[SiliconFly]

.

 

[igor_kavinski]

Buy New ASUS AI Laptop|Discover the best OLED Laptops at Affordable Prices| Now available at ASUS Store

Powered by AI Experience unparalleled performance on Windows 11 Home. Featuring an ultra-fast Intel® Core™ Ultra 7 Series 2 processor with a dedicated AI NPU for 47 TOPs and Intel® Arc™ graphics, the S 14 performs at a higher efficiency and accelerates local AI processes like never before...

shop.asus.com

At least they have priced the 32GB laptop relatively reasonably.

 

[SiliconFly]

.

 

[igor_kavinski]

Yikes! You think 1,500/- for a Lunar Lake laptop is reasonable? Sounds a bit expensive.

Better than what Dell is charging for their 16GB slower one.

And ASUS will discount it after a year or so (if they overproduce )

I see them selling old stock Tiger Lake at below $200 on their local online store (UAE).

 

[SiliconFly]

.

 

[igor_kavinski]

Tiger Lake for 200$ is still a very good deal imho. Pretty decent performance.

I was MOST tempted to get it, but then I already have a "Ice Lake laptop turned mini desktop" that I bought from a 15 year old kid, one 17 inch HP Tiger Lake 32GB RAM laptop that I use for Teams meetings and one Ice Lake 8GB LPDDR4 laptop because the price was too good at something like $190 (plus it came with a stylus!).

 

[511]

R24 number from core ultra 9