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.

 

[ondma]

Yeah may not. But Zen 5 is a flop in uplift with almost no gain in gaming and other consumer workloads that do not use AVX512. Will Arrow Lake actually have a meaningful gain over RPL in gaming and other general consumer workloads even if only 10-15%. I really want to see that plus much lower power for cooler running CPUs and oh stability stability stability which Raptor Lake lacks.

Raptor Lake may even lack stability and still degrade too easily even with corrected microcode and default Intel power profile updates. It was a rushed design and flawed afterall.

"Only" 10-15%? I would jump for joy of ARL shows that much gain in either 1T or nT. (I mean in final performance, not just IPC)

 

[511]

Maybe in 128-bit support, but that's a stretch. Supporting it itself increases transistor count and power because of the extra registers.

Nope Avx10 128 bit was canned so it is 256 bit or 512 bit

AVX512 is a remnant of the misguided and fat Intel that aimed at continually increasing vector performance of CPUs to limit GPU inroads. There were AVX1024 in plans at one point according to some. It should have went from AVX2 to AVX3, but with 256-bit vector.

It is really nice but I agree AMX was invented for AI instead of AVX-1024 a nice step

The E cores have been closely following ARM cores in development. After all, Atom was shifted to compete with ARM in Smartphones early on and that never stopped.

And it is paying them off nicely look how E cores have been constantly getting better i think Skymont is on same level as X2 which is nice in terms of IPC

 

[ondma]

Spotted this. Just one benchmark, but it shows the U7 258V is fallible..

Intel Core Ultra 7 258V beats AMD Z1 Extreme in DX12 gaming benchmark [Updated]

Lunar Lake comes out on top.

www.tomshardware.com

Ashes is a notoriously non-representative "gaming" benchmark. Also, reading the details, there wasnt even a direct comparison in comparable systems.

 

[SiliconFly]

.

 

[Nothingness]

Yes.

Link: C&C

From your link:

AVX-512 is used in Cinebench 2024, but in such low amounts that it’s irrelevant.

So no

 

[MS_AT]

Yes.

Link: C&C

In addition to what Nothingness said I will also quote from the article:

The most common FP/vector math instructions are VMULSS (scalar FP32 multiply) and VADDSS (scalar FP32 add). About 6.8% of instructions do math on 128-bit vectors. 256-bit vectors are nearly absent, but AVX isn’t just about vector length. It provides non-destructive three operand instruction formats, and Cinebench leverages that.

 

[naukkis]

I would buy a 10-12 P core 5GHz Golden Cove/Alder Lake CPU on a single ring bus in a heartbeat if it existed.

Arl come very close of that. Up to 12 threads every thread have 2MB L2 and at least 4.6Ghz clock on same L3 ring. Replacing Skymont clusters with Lion Coves wouldn't change much at sane power levels.

 

[SiliconFly]

.

 

[SiliconFly]

.

 

[TESKATLIPOKA]

No it's just what to be expected from max 17w soc. Those 700-800 results would have only possible without powerlimits consuming about 100w.

As you said, 700-800 points is impossible at low TDP.

LNL is still just a 4P+4E CPU without SMT at 15W+2W for memory, 479 points is not that bad even If some think so or simply had too high expectations.

In comparison:
Ultra 5 125H 4P+8E+2LPE(18t) manages 564 points at 30W(Link).
Ultra 7 164U 2P+8E+2LPE(16t) manages 420 points at 16W(Link).
Core i3-N305 8E(8t) manages 324 points at 15W(Link).

I think retail product will score a bit higher ~500 points.

It could be an interesting CPU for a handheld, but I don't think price will be low and then GPU drivers are still worse than AMDs, and Intel doesn't have any frame generator yet.

 

[Nothingness]

Oh pls. First you said it’s not there at all.

But now you say it’s there but useless.

Make up your mind!

I didn't say anything 😉

Anyway the point you were trying to make that to see AVX-512 benefit one should run Cinebench was utterly wrong. Being wrong is not an issue, insisting on being wrong on the other hand...

 

[Nothingness]

So, like you and this article says, Cinebench leverages AVX512?

Using instructions doesn't mean it's of any use. The graph proves it: no significant amount of AVX-512 is run by CB, so using it to asses the benefit of AVX-512 is meaningless.

 

[511]

As you said, 700-800 points is impossible at low TDP.

LNL is still just a 4P+4E CPU without SMT at 15W+2W for memory, 479 points is not that bad even If some think so or simply had too high expectations.

I agree i was always thinking of 500-550 range

It could be an interesting CPU for a handheld, but I don't think price will be low and then GPU drivers are still worse than AMDs, and Intel doesn't have any frame generator yet.

Funnily amd FG works with Xess so it won't matter much the drivers are the biggest pain though

 

[Abwx]

Oh pls. First you said it’s not there at all.

But now you say it’s there but useless.

Make up your mind!

There s only up to SSE 4.2 in CB R15, a tiny amount of AVX128 in CB R20/23, AVX256 is found only in CB 2024 since this is required to run this bench, and of course there s no AVX512 in whatever version.

 

[Josh128]

So, like you and this article says, Cinebench leverages AVX512?

No, it doesnt. Look at the chart.

 

[SiliconFly]

.

 

[Thunder 57]

Yes.

Link: C&C

What are you talking about? There is no blue (Cinebench) for AVX512.

 

[Thunder 57]

Looks like AVX512 is not useful for general consumers after all!

Transcoding and emulation come to mind.

 

[SiliconFly]

.

 

[Thunder 57]

Only some A/V encoders.

So it is sometimes useful and not never useful?

 

[Tup3x]

Transcoding and emulation come to mind.

Normal user doesn't really do neither though. If normal user does transcoding or video encoding, it's probably done using hardware acceleration.

 

[Nothingness]

Looks like AVX512 is not useful for general consumers after all!

No, it just means the CB guys didn't optimize their software for AVX-512 or found it didn't fit well in their current implementation. I will let you read about the benefits of AVX-512 for some other renderers: https://www.phoronix.com/review/amd-zen5-avx-512-9950x/3

Of course rendering is not of use to all people. Just like high-end gaming isn't.

 

[Markfw]

Transcoding and emulation come to mind.

scientific calculations, primegrid is an example.

 

[gdansk]

It should also work pretty well for a renderer in general.

For example, Intel added it to Embree. But last I tried I couldn't get AVX 512 to work, it required Intel's compiler on Windows. But maybe it works these days.

Not using AVX-512 extensively in a renderer used as a benchmark is almost a deliberate decision. Edit: And it's a fine decision in my opinion.

 

[SiliconFly]

.