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

[Tigerick]

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.



Comparison of upcoming Intel's U-series CPU: Core Ultra 100U, Lunar Lake and Panther Lake

Model	Code-Name	Date	TDP	Node	Tiles	Main Tile	CPU	LP E-Core	LLC	GPU	Xe-cores
Core Ultra 100U	Meteor Lake	Q4 2023	15 - 57 W	Intel 4 + N5 + N6	4	tCPU	2P + 8E	2	12 MB	Intel Graphics	4
?	Lunar Lake	Q4 2024	17 - 30 W	N3B + N6	2	CPU + GPU & IMC	4P + 4E	0	12 MB	Arc	8
?	Panther Lake	Q1 2026 ?	?	Intel 18A + N3E	3	CPU + MC	4P + 8E	4	?	Arc	12

Comparison of die size of Each Tile of Meteor Lake, Arrow Lake, Lunar Lake and Panther Lake

	Meteor Lake	Arrow Lake (20A)	Arrow Lake (N3B)	Lunar Lake	Panther Lake
Platform	Mobile H/U Only	Desktop Only	Desktop & Mobile H&HX	Mobile U Only	Mobile H
Process Node	Intel 4	Intel 20A	TSMC N3B	TSMC N3B	Intel 18A
Date	Q4 2023	Q1 2025 ?	Desktop-Q4-2024 H&HX-Q1-2025	Q4 2024	Q1 2026 ?
Full Die	6P + 8P	6P + 8E ?	8P + 16E	4P + 4E	4P + 8E
LLC	24 MB	24 MB ?	36 MB ?	12 MB	?
tCPU	66.48
tGPU	44.45
SoC	96.77
IOE	44.45
Total	252.15

Intel Core Ultra 100 - Meteor Lake



As mentioned by Tomshardware, TSMC will manufacture the I/O, SoC, and GPU tiles. That means Intel will manufacture only the CPU and Foveros tiles. (Notably, Intel calls the I/O tile an 'I/O Expander,' hence the IOE moniker.)

 

[Magio]

David Huang already has microbenchmarked everything he will release the data on 3rd

That will be very interesting to see but I maintain that until we've seen what the OEMs all do with it we won't really be able to judge LNL's appeal to the consumer.

For my part I'm already sad that for its first year on the market LNL will likely be stuck with the current godawful iteration of the XPS 13, on the Dell side.

 

[AcrosTinus]

A person delivering a message and the content of the message and two different things and even a pathological liar sometimes tells the truth.

So you automatically presuming that the information coming from the source you do not like is false is simply an incorrect inference.

Sorry, Leakers are like a religion, you either believe the crap or not. I will stand on the other side, still you opinion is valid. He might be right, he might be wrong.

For me how you come to the message and how you deal with being wrong shows character. Simply hearing X will be faster than Y or X is cancelled, trust me bro, I heard it from disgruntled and laid off employee XX9783XX, is not enough.

 

[maddie]

Capacity will come, or at least the plans for that aren't kept secret:

Intel 18A and Intel 3 should remain attractive nodes for a good while (including their refinements 18A-P, Intel 3T/E/PT), let's not forget that even today the best chips from Qualcomm, AMD and Nvidia are still on TSMC's 4nm processes and those are very lucrative contracts for TSMC. So if they can ramp the capacity on those as planned, there will be contracts to be found aplenty.

If those are rolled out without many issues, despite the initially limited capacity I can only imagine the prospect of uncontested leadership with 14A would attract extremely lucrative contracts. Of course, great nodes would also be a boon for the product side of the company albeit Intel will have to be careful not to withhold too much of their capacity for themselves if they want to attract the biggest guns in the chipmaking industry.

Of course I'm not saying any of that will come to be, I'm no sightseer. But the plan is there and it is extremely promising on paper.

You wrote, Capacity will come, or at least the plans for that aren't kept secret:

Where exactly? There is no Y axis values. Without those details these slides are pretty worthless. Have to say however, their smoke and mirrors act is not bad. Maybe it'll buy enough time for some recovery.

 

[SiliconFly]

You wrote, Capacity will come, or at least the plans for that aren't kept secret:

Where exactly? There is no Y axis values. Without those details these slides are pretty worthless. Have to say however, their smoke and mirrors act is not bad. Maybe it'll buy enough time for some recovery.

Simple. Just calculate using the scale.

 

[Magio]

You wrote, Capacity will come, or at least the plans for that aren't kept secret:

Where exactly? There is no Y axis values. Without those details these slides are pretty worthless. Have to say however, their smoke and mirrors act is not bad. Maybe it'll buy enough time for some recovery.

There's the current Intel 7/10nm capacity, which is a lot of wafers, as a point of reference and it projects Intel 3/18A matching that towards the end of 2026 and general EUV capacity dwarfing Intel's Foundry's entire current output by the end of the graph.

I'll give you that the advanced packaging graph is completely useless, tho. The starting value is near zero so who the hell knows what the end value is supposed to be !

 

[511]

That will be very interesting to see but I maintain that until we've seen what the OEMs all do with it we won't really be able to judge LNL's appeal to the consumer.

For my part I'm already sad that for its first year on the market LNL will likely be stuck with the current godawful iteration of the XPS 13, on the Dell side.

Well OEM can screw good products really bad i hope Intel puts a little tight leash on evo with LNL

 

[Thunder 57]

You wrote, Capacity will come, or at least the plans for that aren't kept secret:

Where exactly? There is no Y axis values. Without those details these slides are pretty worthless. Have to say however, their smoke and mirrors act is not bad. Maybe it'll buy enough time for some recovery.

So we will see 14A next year? Not buying it. When was that slide released?

 

[Magio]

So we will see 14A next year? Not buying it. When was that slide released?

14A would only be starting risk production in early 2026, it's not expected in HVM until more than a year after that.

The graph is from their February '24 foundry event.

 

[Thunder 57]

14A would only be starting risk production in early 2026, it's not expected in HVM until more than a year after that.

The graph is from their February '24 foundry event.

I get that, since there clearly wasn't products available on 20A/18A in 2023. Only later this year will we see 20A and rumor is ARL will use N3B for the high end and 20A for the low end. They have to really do well with 18A as they seem to be betting their future on it to give me confidence they will return to process leadership.

 

[inquiss]

Maybe but the fool couldn't figure out if Jim Keller was the one to listen to rather than the imposter Renduchintala so Keller left out of exasperation. With Keller advising Pat, things could have turned out a LOT different.

Jim Keller may not have been having the best time because of the infighting etc but he didn't leave intel because of that or the project.

 

[inquiss]

Intel has always been a microprocessor-first company.

Except, you know, when they weren't. They were a memory company first.

 

[Magio]

I get that, since there clearly wasn't products available on 20A/18A in 2023. Only later this year will we see 20A and rumor is ARL will use N3B for the high end and 20A for the low end. They have to really do well with 18A as they seem to be betting their future on it to give me confidence they will return to process leadership.

To your initial point I do think it's likely the 14A timeline is pushed back a little bit, if only because ASML's High NA EUV machines haven't been shipped at quite the rate that was projected then, as far as I can tell.

 

[igor_kavinski]

Isn’t it ironic that Jim Keller now does AI accelerators driven by moderately performing and classical RISC-V cores?

I think he is confident that he can gradually take RISC-V to where others can only dream.

 

[igor_kavinski]

The real positive from the video is that since it's MLID saying it you can safely assume it's entirely made up.

But if there is ANY modicum of truth to that video and Pat really did cancel such an ambitious product, he's dumber than I could have ever imagined. Lofty ambitions are like the fuel of excellent engineers. Wanting to make the impossible possible and watching their baby run millions of devices in the world is what drives them, motivates them and gets them up in the morning with excitement and vigor.

It kinda reminds me of older people looking at younger studs brimming with testosterone and shaking their heads and going, "rabid horny kids". Just because YOU don't feel any excitement in your blood anymore due to lack of testosterone, doesn't mean that you should feel that others WITH testosterone are over-reacting or something. You were like that once too. Remember and remember HARD, you lazy old fart!

 

[igor_kavinski]

Jim Keller may not have been having the best time because of the infighting etc but he didn't leave intel because of that or the project.

Please don't say that he left because of family tragedy or whatever. He got fed up so much by the politics at Intel that he probably swore never to work under anyone again and that's why he started his own company. And this is a guy who has a track record of success at every job in his past. Intel's politics must have been uniquely horrible and unprecedentedly traumatic for him to take such a drastic step as to become his own boss.

 

[jdubs03]

AGAIN, Royal Core is not real, it is not a product. You cannot cancel something that never was. This is a normal R&D exercise in the labs. After things are deemed viable or not, aspects are extracted and used in other products.

How can people be this gullible?
He is basically extracting you for views, feeding you stories that meet your interests, he could basically make things up and people will listen to him.

In reality this video if from a guy that is known to fabricate stuff, claiming to allegedly be talking to a unverified Intel source that somehow spilled the beans to a anti Intel youtuber.

No wonder media literacy is low and idiots can influence stock markets. If fairy tails are given the same weight as researched outlets we are doomed, RIP AnandTech. You inspired me to go through electrical engineering much love.

It’s in a way entertainment. I don’t hold his words as gospel as you seem to think I do. I posted it and it started a discussion about the concept itself being discussed. Also it’s related to other posts about the rumor.

Also my comment after that provides context in which lends some credence to the idea of this project, and its cancellation. Yes it’s a rumor, but I don’t think it is completely made up. It’s been referenced by multiple people that are other than MILD.

 

[SiliconFly]

Please don't say that he left because of family tragedy or whatever. He got fed up so much by the politics at Intel that he probably swore never to work under anyone again and that's why he started his own company. And this is a guy who has a track record of success at every job in his past. Intel's politics must have been uniquely horrible and unprecedentedly traumatic for him to take such a drastic step as to become his own boss.

It's common knowledge Jim Keller had a big fallout with Dr. Murthy Renduchintala during their time at Intel. But instead of throwing the useless Murthy out, Intel as expected chose to keep the nincompoop Murthy and instead let go of Keller. Big mistake.

Intel is still trying to recover from that debacle.

Murthy was fired later for even bigger failures.

 

[igor_kavinski]

Murthy was fired later for even bigger failures.

At the same time, Intel should have sent an apology letter to Keller.

But of course, if anyone there had the slightest courage to admit their mistakes, they would not have landed their entire company in trouble ever.

 

[SiliconFly]

It’s in a way entertainment. I don’t hold his words as gospel as you seem to think I do. I posted it and it started a discussion about the concept itself being discussed. Also it’s related to other posts about the rumor.

Also my comment after that provides context in which lends some credence to the idea of this project, and its cancellation. Yes it’s a rumor, but I don’t think it is completely made up. It’s been referenced by multiple people that are other than MILD.

Agree. Even though he thrives on rumors and made up stories most of the time, on rare occasions, he does come up with some unexpected news. He was the first one to report that Raja and AXG are getting the axe (split & reorg). Everyone thought he was lying, but it actually happened around 6 to ~9 months later. Also, he was the first one to report about Rentable Units. On the other hand, his understanding of tech is so weak, it's safe to say he's almost a tech illiterate. But it's people like him who keep things interesting. He's good entertainment.

 

[DavidC1]

It is absurd to think that getting your ideas shot down is the motivating reason why you would leave a company, especially when the company is in trouble as is.

That's not why they said the team members left. Intel put them in charge of "AIeee" divisions instead. It must have been like a slap in the face to them. To go from their dream project to what is essentially engineer FOMO.

 

[inquiss]

Please don't say that he left because of family tragedy or whatever. He got fed up so much by the politics at Intel that he probably swore never to work under anyone again and that's why he started his own company. And this is a guy who has a track record of success at every job in his past. Intel's politics must have been uniquely horrible and unprecedentedly traumatic for him to take such a drastic step as to become his own boss.

He took time out for family tradgedy yes - and didn't know how long that was going to be for. So he had to step back. Then why go back to intel? Toxic for the exact reasons you mentioned.

 

[DavidC1]

He took time out for family tradgedy yes - and didn't know how long that was going to be for. So he had to step back. Then why go back to intel? Toxic for the exact reasons you mentioned.

Jim Keller leaving Intel for family tragedy is the same excuse that Brian Kraznich left for "sexual relations with an employee".

No Board of Director cares in 2000's America about sexual relations with an employee. They get away with MUCH bigger than that. There was a story about Broadcom CEO engaging in sex parties in his own shack not far away from his home before his wife found about it.

What about allegations about "secret parties" given to VIP attendees after Intel Developer Forum?

You have to read between the lines. It's better to basically treat any press release as lies. Basically they wanted to fire those people but didn't have a politically correct way of firing them. So for example when BK engaged in "it" it gave them an excuse. The real reason he was fired was because he was a terrible CEO and led Intel to doom.

 

[DavidC1]

I get that, since there clearly wasn't products available on 20A/18A in 2023. Only later this year will we see 20A and rumor is ARL will use N3B for the high end and 20A for the low end. They have to really do well with 18A as they seem to be betting their future on it to give me confidence they will return to process leadership.

So basically you have to shift the timeframe 2 years into the future. Meaning 2026 is really 2028. So they are basically admitting no real volume until 2028.

You can verify that by the fact that the 20/18A volume ramp doesn't really go up until 2024-2025(really 2026-2027), which you cannot have since it's almost 2025 and none of them really exist. 20A is nonexistent and is going to be nonexistent, so it's with Pantherlake in 2026 when the ramp happens. See? 2 years again.

Pantherlake probably gets introduced in late 2025, meaning volume ramp has to happen few quarters before. But it's actual selling that really ramps up the volume. Shortly after is Clearwater Forest.

14A ain't happening until 2 years after 18A, or 2027. Intel said they are going back to the "traditional cadence". Too bad the performance gains are 3/4 node and density gains are a 1/4 node. Explains why we're going to have bunch of -P and -T nodes.

 

[Philste]

I'm not sure Royal Core was ever real. Also these MLID Leak from today is so weird again. So he really wants to tell us that Intels Flagship 2028 product would've been 6(!) Cores?!? Even with the claimed double IPC that would mean Intels 2028 Product would have same MT than their 2022 product, that makes no sense. 6 of those Cores would never be a highend part in 4 years, even if they can split up to 24 Threads.

 

[DavidC1]

I'm not sure Royal Core was ever real. Also these MLID Leak from today is so weird again.

Please tell the AheadComputing team that they don't exist.

I'm guessing needing to be shown in front of their face is why some don't believe in the Moon Landing for example. I thought of it myself too. Like what if ALL airplanes do is that it stays in one place, but it's the scenery that changes instead?

So he really wants to tell us that Intels Flagship 2028 product would've been 6(!) Cores?!? Even with the claimed double IPC that would mean Intels 2028 Product would have same MT than their 2022 product, that makes no sense. 6 of those Cores would never be a highend part in 4 years, even if they can split up to 24 Threads.

Royal Core version 1 was supposed to be a 24-wide issue one with novel ideas like Value Prediction and 2x perf/clock over Raptor Cove. The version 2 was supposed to be 3x.

6 cores is ok with 2x Raptor Cove. It would have accelerated low thread workloads like a F1 racing car, and when MT is needed it would have quadrupled threads - according to MLID anyways.