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.

 

[ajsdkflsdjfio]

If they had been so good, they could've put in the extra effort to keep AVX-512. Haven't heard of Zen4c/5c?

Maybe AVX-512 isn't so important for 99% of users yet?

 

[Win2012R2]

Maybe AVX-512 isn't so important for 99% of users yet?

It wasn't important thanks to Intel's decision to keep it server only, without installment base on consumers side nobody will optimise for it and server side Intel's implementation in Skylake was so terrible general consensus was that outside very specific HPC it was best not to touch it with a bargepole. But this all changed with Zen 4, it will still takes years but should be a lot quicker now

 

[ajsdkflsdjfio]

It wasn't important thanks to Intel's decision to keep it server only, without installment base on consumers side nobody will optimise for it and server side Intel's implementation in Skylake was so terrible general consensus was that outside very specific HPC it was best not to touch it with a bargepole. But this all changed with Zen 4, it will still takes years but should be a lot quicker now

Sure but from Intel's POV it might not be beneficial to put in the extra effort to make AVX-512 work even if "they had been so good". Like you said, it will take years even with Zen4 2x256 AVX--512 and ZEN5 full AVX-512 support. Even if AVX-512 adoption was instant, if Intel dictates the market, leaving out AVX-512 makes AVX-512 irrelevant making their decision to remove AVX-512 non-detrimental to their products' competitiveness.

Also the comparison of ATOM vs Zen C cores isn't really fair. ZenC uses the same core architecture as regular Zen cores and is around 25% smaller. Atom cores are 1/4-1/3 of the size of P-cores and use an entirely separate architecture. They are designed for different things, it's not simply that Intel's engineering team "wasn't good enough" that they left out AVX-512 on their E-cores. In the first place, ZenC cores aren't even made for regular consumers, I'm pretty sure they only exist on like 1 or two Zen products in laptop.

I understand the sentiment that Intel may have slowed down innovation for consumers, but from a business perspective it makes perfect sense as Zen4/5 pays the early adopter fees of including AVX-512 in their client architectures while Intel only spends the extra effort for AVX-512 in consumer products when it actually matters.

 

[igor_kavinski]

Maybe AVX-512 isn't so important for 99% of users yet?

If you build it, they will come. But Intel built it half-heartedly and then outright disabled it in consumer space. Probably after Zen 6 or Zen 7, there should be enough installed consumer base of AVX-512 enabled Ryzen CPUs for developers to start caring. Or maybe AVX10 will fix the lack of developer attention towards optimizing the important bits in their codebase to get the full benefit of everything the actual inventors of AVX-512 envisioned.

 

[Win2012R2]

from Intel's POV it might not be beneficial

From Intel's point of view it wasn't beneficial to sell 8 cores instead of 4, for the same price.

AVX-512 isn't just about double size registers, it has got very useful instructions that can run at full speed even on Zen-4 - they help speed up a lot of workloads that got nothing to with HPC double floats even.

 

[igor_kavinski]

In the first place, ZenC cores aren't even made for regular consumers, I'm pretty sure they only exist on like 1 or two Zen products in laptop.

It's part of their Zen 5 mobile line-up.

 

[igor_kavinski]

From Intel's point of view it wasn't beneficial to sell 8 cores instead of 4, for the same price

Yeah. If they could have it their way, they would keep screwing over consumers till the end of times.

AMD is the equalizer.

 

[ajsdkflsdjfio]

From Intel's point of view it wasn't beneficial to sell 8 cores instead of 4, for the same price

And they got away with it for a long time lol. Except that they also let their engineering teams and technology fall behind which lead them to now.

View attachment 113959
It's part of their Zen 5 mobile line-up.

That's what I meant by 1 or 2 mobile products?? It's not in 90% of their product stack like Intel E-cores are.

 

[ajsdkflsdjfio]

Yeah. If they could have it their way, they would keep screwing over consumers till the end of times.

AMD is the equalizer.

I don't disagree, why are you making a point of this? My only point was to say that it's not that Intel's engineering team simply isn't "good enough" to put in the extra effort to add in AVX-512 to e-cores.

You think Nvidia making much of their lower end lineup have low amounts of VRAM compared to AMD is because their engineering team isn't good enough? No, they just understand that it won't hurt their sales or profits and choose to make decisions worse for consumers.

 

[Win2012R2]

I've been keeping an eye on AVX-512 dev since it was out and recent couple of years show marked spike in developments - massive speedups in areas where one would never think it would be useful, this isn't going away

 

[ajsdkflsdjfio]

I've been keeping an eye on AVX-512 dev since it was out and recent couple of years show marked spike in developments - massive speedups in areas where one would never think it would be useful, this isn't going away

I agree? Notice how I included "yet" in my original statement.

 

[Win2012R2]

Intel's engineering team simply isn't "good enough"

Their initial implementation for servers was so terrible it put people off using AVX-512 for many years, it has gained such a bad notoriety that its usage was discouraged

 

[igor_kavinski]

It's not in 90% of their product stack like Intel E-cores are.

They aren't as dumb as Intel. Or out of options due to process related issues.

 

[ajsdkflsdjfio]

Their initial implementation for servers was so terrible it put people off using AVX-512 for many years, it has gained such a bad notoriety that its usage was discouraged

What does that have to do with putting AVX-512 in e-cores? It's an entirely separate team and entirely different goals.

They aren't as dumb as Intel. Or out of options due to process related issues.

Yes they are in a corner because of poor past decisions, doesn't make the E-cores themselves a bad bet. If you shoot yourself in the leg and bandage your wound later on, it's not the bandage that's the issue it's being shot in the leg.

Putting E-cores in their products isn't dumb, it's the only way they've been able to feign competitiveness for the past 4 generations of CPUs.

 

[SiliconFly]

.

 

[Win2012R2]

It's better to go with AVX10 instead of the messy AVX-512 for clients

CPUs supporting AVX10 also support legacy AVX-512 stuff, the new /256 things require recompile though, this will certainly help AVX-512 adoption, but I doubt AVX-10 specifics will be that popular, AMD has got far smarter "256" implementation that Intel should have copied, perhaps that will happen once they trash out details in their new "advisory" group

AMD managed to capture only around ~20% client market share.

AMD had and still has limited capacity so they focus on most profitable markets, plus Intel pays heavily OEMs like Dell to keep market share and they owned laptops for super long time, very hard to break in.

 

[igor_kavinski]

My only point was to say that it's not that Intel's engineering team simply isn't "good enough" to put in the extra effort to add in AVX-512 to e-cores.

They proudly state that they have world-class engineers. They made those engineers work on AVX-512 and validate it for Alder Lake. Then at the last moment (probably idiot Pat's doing), they disabled it, much to the utter surprise and chagrin of those very engineers. This is what you get when you have competing teams in the same organization. The E-cores could've had AVX-512 if the relevant experienced engineers had been tapped to offer their expertise to help the E-core teams.

Intel simply embarked on a blind path towards regaining their lost mindshare and kept making missteps over and over till Pat finally couldn't keep himself from getting fired. I hope to dear God that with AVX10, they finally make up their minds not to ditch what they start.

 

[ajsdkflsdjfio]

I really don't get the debate here? I agree with 90% of what you two are saying about AVX-512 usefulness and how Intel slowed down the adopted of AVX-512 in the consumer market.

I simply believe that the decisions to not have AVX-512 in e-cores doesn't affect their product competitiveness and therefore wasn't just because the engineering teams behind the ATOM cores weren't clever enough.

 

[igor_kavinski]

Lol. Years of Intel failure and AMD managed to capture only around ~20% client market share.

Because Intel is the master of marketing deception. That didn't work so well with Raptor Lake Refresh and Arrow Lake though. 9800X3D is selling like hot cakes. Where is Intel's answer to that?

 

[ajsdkflsdjfio]

They proudly state that they have world-class engineers. They made those engineers work on AVX-512 and validate it for Alder Lake. Then at the last moment (probably idiot Pat's doing), they disabled it, much to the utter surprise and chagrin of those very engineers. This is what you get when you have competing teams in the same organization. The E-cores could've had AVX-512 if the relevant experienced engineers had been tapped to offer their expertise to help the E-core teams.

Intel simply embarked on a blind path towards regaining their lost mindshare and kept making missteps over and over till Pat finally couldn't keep himself from getting fired. I hope to dear God that with AVX10, they finally make up their minds not to ditch what they start.

The "probably idiot Pat's doing" is just massive cope lol. How could you know that? And anyways as multiple people here have stated in the past, AVX-512 for Alderlake and subsequent generations didn't work because of architectural problems NOT because they decided to turn it off for no good reason. And the decision to not have AVX-512 in e-cores LONG predates Pat so I don't understand where you get the idea that AVX-512 not being turned on in Alder-lake was Pat's fault.

 

[Win2012R2]

Then at the last moment (probably idiot Pat's doing), they disabled it

Well, they did not really have a choice - their implementation is super inefficient anyway, hitting clocks too much, way too much heat and that's in super highly clocked CPU - plus they'd have to only offer 8 cores that were expensive to make, I would not blame Pat for that.

Because Intel is the master of marketing deception

It's them "co-marketing" funding, but the money drying up, they can't afford to essentially bribe OEMs in server space no more.

 

[SiliconFly]

.

 

[ajsdkflsdjfio]

Because Intel is the master of marketing deception. That didn't work so well with Raptor Lake Refresh and Arrow Lake though. 9800X3D is selling like hot cakes. Where is Intel's answer to that?

Cope. I agree AMD should have more marketshare given a more flexible market, but consumers are motivated to stay towards tradition. It's not that Intel is extra deceptive that they are able to keep marketshare, it's that consumers aren't so quick to switch over to a new side.

AMD is plenty deceptive in their marketing, I mean have you seen their Zen5 release day numbers? Did you not realize that their Zen5 "efficiency" gains were simply running their X-series cpus at non-X TDP and then comparing those numbers to Zen4-X CPUs using 50% more power? They were called out and received massive backlash and then said "oopsies" then increased Zen5-X power limits back to normal, eliminating any so called "efficiency" gains.

All these companies are masters of deception it's just the nature of the game. If anybody would be a master of deception it would be Nvidia, but alas they also genuinely have the advantage anyways.

 

[igor_kavinski]

I simply believe that the decisions to not have AVX-512 in e-cores doesn't affect their product competitiveness and therefore wasn't just because the engineering teams behind the ATOM cores weren't clever enough.

They might be clever enough but I was referring more to Intel's "excuse" that AVX-512 can't work with E-cores enabled. A hardware/software combination of "trap AVX-512 instruction and context switch over to P-core" could've been done if they hadn't been lazy and so sure of themselves.

 

[Win2012R2]

Nope. They aren't backward/forward compatible. AVX-512 code can't just be recompiled to work with AVX-10 and vice versa. They need to be recoded. Time to say goodbye to AMD's client implementation of AVX-512.

All "legacy" AVX-512 will run on AVX-10, however in order to take advantage of new stuff (like half speed AVX-10/256) it will require recompile (from what I've read) or maybe minor recode, very few people will be doing it and certainly nobody is going to be disabling working AVX-512 path in favour of zero installment AVX-10 base.

So AMD's approach is the best - I expect Intel will copy it too as it will be easy with AVX-10/256 to map stuff and double pump it, that's the only way it will work in "e-cores", but if they don't then good luck to them waiting for support

A hardware/software combination of "trap AVX-512 instruction and context switch over to P-core" could've been done if they hadn't been lazy and so sure of themselves.

Nah, that's deadend - plus they had no time to implement it as it would certainly require hardware support, disabling was the only sensible option