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.

 

[AMDK11]

No, the claim is that 14th gen desktop will be Raptor Lake refresh, but that MTL-S will also exist, and arrive alongside Arrow Lake, likely on the low end.

This makes sense because if the Meteorlake-S is temporally close to the ArrowLake-S, the Meteor can easily be considered low-end models with fewer cores and lower performance. Especially since ArrowLake is on the brand new LionCove x86 cores with Wide 8 and ROB 700+. if you believe the leaks.

 

[A///]

did royal core get shelved or is there something being done for it?

They're the same root. I've heard both used.

Intel's lost many employees since Meteor Lake started. They're everywhere now, and some are willing to talk about little details with their new coworkers.

one of them only shows comic book and comic book film sources online. perhaps you're spelling it wrong? that's what I though those would be during the murthy era firings right? i don't remember if there were any mass layoffs during the swan era, but I remember at the end of his tenure swan and then gelsinger began to hire back some of the bright engineers that had retired or had been let go.

 

[igor_kavinski]

Maybe Intel will try to mimic AMD's strategy of having some OK performing APUs with better performing iGPUs based on MTL and flagship CPUs based on ARL.

 

[Exist50]

one of them only shows comic book and comic book film sources online. perhaps you're spelling it wrong?

I'm not 100% sure what the true name is. Seems like the abbreviation is used so frequently instead that even Intel folk aren't quite sure. "Adamanium" would make more sense, given that it's a noun vs the adjective "adamantine", but the latter we actually have proof of, so... But I don't think the comic book/fantasy relation would be a sticking point. Would hardly be out of the ordinary for a tech company. Anyway, just semantics at the end of the day.

that's what I though those would be during the murthy era firings right? i don't remember if there were any mass layoffs during the swan era, but I remember at the end of his tenure swan and then gelsinger began to hire back some of the bright engineers that had retired or had been let go.

I think they had some small layoffs under Swan, but I'm talking about predominantly voluntary departures. As I've said before, Meteor Lake was extremely poorly managed. When a program is mismanaged, it both puts pressure on the workers (increased workload), and demoralizes them (no one likes seeing their work get thrown out because of a definition, or to feel like no end is in sight). This drives people away, especially at a company like Intel that pays below market rate. Add in some poaching, a sprinkle of layoffs, and tada - you've got a whole bunch of people eager to share their war stories with a sympathetic ear.

Anyway, if Gelsinger has brought back some people, it doesn't seem to have made much of an improvement to their client teams. Everything I've heard seems like they're still flying by the seat of their pants. Seems like most of the effort has gone into shoring up the manufacturing teams and server.

 

[igor_kavinski]

This drives people away, especially at a company like Intel that pays below market rate.

That's still 120K, right? RIGHT?

 

[A///]

I'm not 100% sure what the true name is. Seems like the abbreviation is used so frequently instead that even Intel folk aren't quite sure. "Adamanium" would make more sense, given that it's a noun vs the adjective "adamantine", but the latter we actually have proof of, so... But I don't think the comic book/fantasy relation would be a sticking point. Would hardly be out of the ordinary for a tech company. Anyway, just semantics at the end of the day.

Adamanium sure but adamantium brings up superhero stuff dating back to the 1960's. including the word intel brings up nothing about the company but does bring up the word intel like intel gathering.

I think they had some small layoffs under Swan, but I'm talking about predominantly voluntary departures. As I've said before, Meteor Lake was extremely poorly managed. When a program is mismanaged, it both puts pressure on the workers (increased workload), and demoralizes them (no one likes seeing their work get thrown out because of a definition, or to feel like no end is in sight). This drives people away, especially at a company like Intel that pays below market rate. Add in some poaching, a sprinkle of layoffs, and tada - you've got a whole bunch of people eager to share their war stories with a sympathetic ear.

Anyway, if Gelsinger has brought back some people, it doesn't seem to have made much of an improvement to their client teams. Everything I've heard seems like they're still flying by the seat of their pants. Seems like most of the effort has gone into shoring up the manufacturing teams and server.

even if Pat "Guns A'blazing Gunslinger Gelsinger" brought back everyone including juniors and the green and seasoned process techs they'd still be screwed six ways from sunday with amd on their back in dc and now mobile where most of their profits are. the diy scene is small and the prebuilt desktop isn't that big and in flux due to the buy outs by companies during the pandemic.

while having lunch earlier today i was looking up the adamantine stuff and came across another or the full uspto listing with more technical material.

Early Platform Hardening Technology For Slimmer And Faster Boot Patent Application

U.S. Patent Application 20210081538 for Early Platform Hardening Technology For Slimmer And Faster Boot

uspto.report

something i often forget about the uspto website is it's name sensitive. Typing out vincent zimmer or his colleagues don't bring up much but if you go with last name first it brings up first name last name results. some very interesting filings below. with a a weekend spent at the computer and looking up filled positions at Intel of people suspected to work on meteor and arrow lake and searching them on the uspto page there'd be some better info to dig through saving time from going through all intel corp filings, most of which would not interest any of us.

Zimmer; Vincent Patent Filings

uspto.report

Zimmer; Vincent J. Patent Filings

uspto.report

 

[A///]

That's still 120K, right? RIGHT?

sure but that number isn't much higher than what I remember them offering me 18 years ago. I left them hanging dry at the 11th hour after another offer came in for a non x86 company but he move would be the south. slightly higher, better benefits for the time and lcol. my then wife was both happier but also sad about it being the south. overall great experience. if that number is base salary then they're offering sweeter pot these days but that has to be total.

I missed out on a period of exponential growth they had thanks to c2d and amd languishing whatever lead they had and continually messing up. I don't care. had it been apple I would have had real regrets and depression by now.

 

[Exist50]

That's still 120K, right? RIGHT?

Idk, but levels.fyi seems to think they indeed offer >120k.
https://www.levels.fyi/companies/intel/salaries

It's just that e.g. Apple and Nvidia offer more. Though I think AMD is pretty similar.

 

[Exist50]

even if Pat "Guns A'blazing Gunslinger Gelsinger" brought back everyone including juniors and the green and seasoned process techs they'd still be screwed six ways from sunday with amd on their back in dc and now mobile where most of their profits are

With today's products, sure, but I'm speaking more about the state of things still unreleased. They seem to be getting themselves together a bit in server, but client is still disorganized.

 

[Hulk]

Back in the tick/tock days Intel would advance generations simply using a die shrink with minimal core changes. Sandy to Ivy Bridge and Haswell to Broadwell come to mind. I have a feeling we're looking at something similar happening here with Raptor Lake to Meteor Lake. MTL and Raptor Refresh will somehow coexist on the desktop as 14th gen perhaps? We have seen mixed cores for generations split between mobile and desktop but never mixed cores on the desktop in the same generation AFAIK. I mean mixed Big cores and not counting the rebranded Alders in Raptor 13th gen.

 

[A///]

With today's products, sure, but I'm speaking more about the state of things still unreleased. They seem to be getting themselves together a bit in server, but client is still disorganized.

yes I'm one of the few impressed by the 3495x which most seem to be dumping on. if the power use was under further control it would be a great processor and still is in its current form for anyone who needs more power than last gen tr and epyc doesn't make sense for their use case. amd's biggest weakness for their ws would be the artificial separation of regular tr and tr pro the latter of which always seemed to be locked behind an oem which only hurts them as those who need its specs will go to intel. performance may only be 90% there with higher power use but hitting the latter in their likely work loads will not be a common issue. the expansion offered on W790 is impressive. z790 expansion isn't so good, but also dubious on x670 due to lacking componentry that doesn't exist at the moe but even id it does it doesn't offer a noticeable increase over mature last gen products

 

[A///]

Back in the tick/tock days Intel would advance generations simply using a die shrink with minimal core changes. Sandy to Ivy Bridge and Haswell to Broadwell come to mind. I have a feeling we're looking at something similar happening here with Raptor Lake to Meteor Lake. MTL and Raptor Refresh will somehow coexist on the desktop as 14th gen perhaps? We have seen mixed cores for generations split between mobile and desktop but never mixed cores on the desktop in the same generation AFAIK. I mean mixed Big cores and not counting the rebranded Alders in Raptor 13th gen.

that does seem to be what Pat "Guns A'Blazin Gunslinger" Gelsinger was hinting at a year ago or longer. the -2 p cores seem to have rattled many ofc.

 

[A///]

Idk, but levels.fyi seems to think they indeed offer >120k.
https://www.levels.fyi/companies/intel/salaries

It's just that e.g. Apple and Nvidia offer more. Though I think AMD is pretty similar.

what is this? Is it like glassdoor? what's the process of verifying some of those eye watering claims?

 

[Exist50]

what is this? Is it like glassdoor? what's the process of verifying some of those eye watering claims?

levels.fyi is kinda the go-to for salary comparisons in the tech industry these days. Not sure how exactly they verify individual submissions, but their averages seem to track well with anecdotal experience across a number of employers, so I think you can trust what you see.

 

[Exist50]

did royal core get shelved or is there something being done for it?

Oh, and as far as I can tell, Royal is still very much alive and well. But when we'll actually see products with it is anyone's guess at this point. I think the latest rumors pointed to Nova Lake, was it? But tbh the rumor mill still doesn't seem to know what Royal even is, so I wouldn't put any real weight on that.

 

[A///]

Oh, and as far as I can tell, Royal is still very much alive and well. But when we'll actually see products with it is anyone's guess at this point. I think the latest rumors pointed to Nova Lake, was it? But tbh the rumor mill still doesn't seem to know what Royal even is, so I wouldn't put any real weight on that.

it's been way too long for me to remember. I remember asking, it being discussed and the conversation moving rapidly towards meteor. Meteor, arrow, lunar for mobile, panther on desktop. I looked up nova lake a couple weeks back but mentions of it ceased to exist after early january 2022, it's taken a firm backseat in everyone's minds especially due to it not being confirmed or any info about it since then. the closest we'll get to it being confirmed is if diagnostics tools update their codebases with prelim data, but those are often done only 4-6 months before a product release. Only post loading of prelim data do leaks begin to appear which are closer to the final product than es leaks far further out which are un optomized and generally say nothing to end users other than it exists and is being tested. bless careless third party companies that do their tests and forget to disconnect.

 

[AMDK11]

With the refreshed RaptorLake and MeteorLake so close, why is there no official word from Intel now?

We constantly rely solely on the speculation and rumors of alleged Intel employees/former employees. How much truth and current state of affairs are there in these leaks/rumours?

I wonder if there are any details about RedwoodCove or just rumors from rumors? Micro-architectural data and what has been rebuilt and extended. Any numbers. do we have anything?

It is 100% certain, judging by the structure from the Meteorlake photos, that RedwoodCove is not the same as RaptorCove. Some changes have taken place and it is difficult to judge from the photos of the structure alone whether these changes are very minor or maybe larger than is commonly speculated.

Can't the RaptorLake refresh be just a nod to the LGA1700? After all, this is the last platform to support DDR4. The new MeteorLake and ArrowLake platform is already on LGA1851 which only supports DDR5.

At this point, against the advice of others, I dare assume that the RaptorLake-S refresh is not instead of MeteorLake-S. Unless Intel officially announces it.

 

[Geddagod]

With the refreshed RaptorLake and MeteorLake so close, why is there no official word from Intel now?

We constantly rely solely on the speculation and rumors of alleged Intel employees/former employees. How much truth and current state of affairs are there in these leaks/rumours?

I wonder if there are any details about RedwoodCove or just rumors from rumors? Micro-architectural data and what has been rebuilt and extended. Any numbers. do we have anything?

It is 100% certain, judging by the structure from the Meteorlake photos, that RedwoodCove is not the same as RaptorCove. Some changes have taken place and it is difficult to judge from the photos of the structure alone whether these changes are very minor or maybe larger than is commonly speculated.

Can't the RaptorLake refresh be just a nod to the LGA1700? After all, this is the last platform to support DDR4. The new MeteorLake and ArrowLake platform is already on LGA1851 which only supports DDR5.

At this point, against the advice of others, I dare assume that the RaptorLake-S refresh is not instead of MeteorLake-S. Unless Intel officially announces it.

Suggested increased L1 cache from the die shot of RWC. L1 cache changes can have a major impact on performance. I think it's going to be another Palm Cove moment, which also saw L1 cache changes on a new node.
Also Intel has an event lined up for the 27th? Some Q1 event or investor meeting, I don't remember. Maybe they announced RPL-R there.

 

[AMDK11]

Yes. From the Meteorlake drop, it was speculated that L1I RedwoodCove is also 48KB like L1D.

Edit:
PalmCove is generally a slightly improved Skylake in 10nm. Support for AVX512 was added, but it was a big compromise and was mostly based on 2x256 bits very similar to Skylake. SunnyCove is just introducing strong FPU expansion. In addition, L/S capabilities, scheduling and minor nuances in the predictor have been slightly enhanced. The changes probably gave a negligible increase in IPC and only upgrading to AVX512 gave any noticeable.

If Meteorlake was going to be after AlderLake, it should be LGA1700 compliant.

I think we should know the truth soon.

 

[SiliconFly]

With the refreshed RaptorLake and MeteorLake so close, why is there no official word from Intel now?

We constantly rely solely on the speculation and rumors of alleged Intel employees/former employees. How much truth and current state of affairs are there in these leaks/rumours?

I wonder if there are any details about RedwoodCove or just rumors from rumors? Micro-architectural data and what has been rebuilt and extended. Any numbers. do we have anything?

It is 100% certain, judging by the structure from the Meteorlake photos, that RedwoodCove is not the same as RaptorCove. Some changes have taken place and it is difficult to judge from the photos of the structure alone whether these changes are very minor or maybe larger than is commonly speculated.

Can't the RaptorLake refresh be just a nod to the LGA1700? After all, this is the last platform to support DDR4. The new MeteorLake and ArrowLake platform is already on LGA1851 which only supports DDR5.

At this point, against the advice of others, I dare assume that the RaptorLake-S refresh is not instead of MeteorLake-S. Unless Intel officially announces it.

It's kinda strange. Long back Intel used to release powerpoints in advance about the products they're working on. Even before it samples. Now they're trying to behave like Apple... everything is shrouded in secrecy! Don't know why the sudden change! Maybe it gives them some advantage. Not sure.

But one thing we can be sure of is, Redwood Cove is not a refresh of Raptor Cove. Judging by the sheer amount of changes from monolithic to dis-aggregated, including tiles, faveros, L4, etc., I'm sure everything has been retooled completely, including the core architectures (both P & E cores).

Also, clock regression shouldn't be that big an issue for MTL as Intel 4 has been around for a while. They even released a specialty processor last year built on Intel 4. And they'ye had enough time to workout the initial performance issues on Intel 4. So, MTL should be able to hit decent clocks compared to previuos gen. But it looks like they're keeping the core count low due to either yield, or power or heat issues. Still no info on that either.

 

[BorisTheBlade82]

SunnyCove is just introducing strong FPU expansion. In addition, L/S capabilities, scheduling and minor nuances in the predictor have been slightly enhanced. The changes probably gave a negligible increase in IPC and only upgrading to AVX512 gave any noticeable.

SNC was very wide compared to SKL, which could be seen by a hefty ROB size increase from 224 to 352, see link below. The IPC gain was also significant with 15-20% IIRC.
The trouble was, that they couldn't get it to clock nearly as high as SKL. This could only be mostly fixed with Tiger Lake and finally Alder Lake.
But the Sunny Cove DNA is still there, even in Raptor Cove. That is a significant reason for their high power draw at most ISO performance levels compared even to Zen3.

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

 

[AMDK11]

Is there a picture of the structure of the Meteorlake 6+16 system? If not, what is the certainty that 8+16 does not exist?

 

[AMDK11]

SNC was very wide compared to SKL, which could be seen by a hefty ROB size increase from 224 to 352, see link below. The IPC gain was also significant with 15-20% IIRC.
The trouble was, that they couldn't get it to clock nearly as high as SKL. This could only be mostly fixed with Tiger Lake and finally Alder Lake.
But the Sunny Cove DNA is still there, even in Raptor Cove. That is a significant reason for their high power draw at most ISO performance levels compared even to Zen3.

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

About SunnyCove I wrote only a mention of a large expansion of the FPU in relation to PalmCove / Skylake. The rest of my post was just about PalmCove in relation to Skylake.

 

[BorisTheBlade82]

@SiliconFly
Might it be, that Intel simply is not as convinced of MTL as to tease and showboat it? And by the way, they already did a big technical deep dive for it last year.
Wasn't it that most sources only talk about minor design changes for the cores in order to not further increase complexity and risk for their product? I mean, tile architecture, new process, foundry partnership on the same SKU already is s lot of potential trouble.
Also, what kind of experience with Intel4. To my knowledge, their was no product even remotely comparable to MTL before on that process. Or do you mean experience by doing revisions? If so, is there a number in the public about how many revisions MTL already had?

 

[BorisTheBlade82]

About SunnyCove I wrote only a mention of a large expansion of the FPU in relation to PalmCove / Skylake. The rest of my post was just about PalmCove in relation to Skylake.

So not the whole paragraph I quoted was about Sunny Cove? If so, you might consider to restructure that a bit IMHO.