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	8 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)	Arrow Lake Refresh (N3B)	Lunar Lake	Panther Lake
Platform	Mobile H/U Only	Desktop Only	Desktop & Mobile H&HX	Desktop Only	Mobile U Only	Mobile H
Process Node	Intel 4	Intel 20A	TSMC N3B	TSMC N3B	TSMC N3B	Intel 18A
Date	Q4 2023	Q1 2025 ?	Desktop-Q4-2024 H&HX-Q1-2025	Q4 2025 ?	Q4 2024	Q1 2026 ?
Full Die	6P + 8P	6P + 8E ?	8P + 16E	8P + 32E	4P + 4E	4P + 8E
LLC	24 MB	24 MB ?	36 MB ?	?	8 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.)

 

[FlameTail]

The difference is that now Microsoft, ARM, AMD, and Nvidia are all actively working together to replace x86. Microsoft is optimizing Windows 12 for ARM--rather than just making a crummy port of Windows to run on a couple obscure laptops. This new collaboration is for laptops: https://www.theverge.com/2023/10/23/23929240/nvidia-amd-cpu-arm-pc-chips-2025-release-rumors
desktop: https://www.microsoft.com/en-us/d/windows-dev-kit-2023/94k0p67w7581?activetab=pivot:overviewtab
and all the way through cloud: https://news.microsoft.com/source/features/ai/in-house-chips-silicon-to-service-to-meet-ai-demand/

As long as people think it is an AMD vs Intel battle, then both companies will lose the war against Microsoft + Nvidia + ARM.

I realize that I'll never convince anyone to give up their favored chip company. But, I think we can at least stop encouraging others to root for "your team" as that is actually harming your team. Both make quite good chips. Far better than anything we had just a few years ago. However, neither company can survive ARM if they fall into petty squabbles.

Down with x86!
Long live ARM!

 

[SiliconFly]

That Network on the chip of Zen 6 will be a server grade variant, based on the Pensando datacenter DPU cards.

But the LP cores for client will feature only in Zen 6. And it'll be sad if Zen 6 client still doesn't support NOC.

 

[FlameTail]

It will be a new design by a defected Intel CPU architect who gets disgusted with whatever is going on at Intel right now. He's too impatient for Nova Lake to arrive in 2026 and he loves a challenge so he's funning around at some competitor's underground lab as we speak, working on pushing ARM to the moon and kick Nova Lake way past Mars.

And who is this architect you speak of?

 

[FlameTail]

A compelling reason for me would be an ARM CPU working at 7 GHz. Even if it runs x86 software in emulation mode, eventually there will be enough native software able to take full advantage of 7 GHz frequency and leave behind similar legacy x86 software in the dust. The higher your CPU goes, the more your mind blows!

The entire philosophy of current ARM cores is to run slow and wide. I am not aware of any ARM core which has a higher clock speed than the X Elite's Oryon @ 4.3 GHz. And even then we had discussed in the Hamoa thread how this core at 4.3 GHz might be pulling a lot of power as we can guess it's way out of it's efficienct range at that clock speed.

 

[SiliconFly]

Down with x86!
Long live ARM!

Down with ARM!
Long live RISC-V!

 

[FlameTail]

Making a world class high performance and efficient ARM laptop/desktop class CPU isn't an issue anymore. Apple has already shown the way. But the big question is, will they succeed? I don't think it's gonna happen any time now.

Microsoft Windows managed to spawn such a massive x86 software ecosystem over the decades, an ARM software ecosystem won't be able to replace it in a blink of an eye. It's gonna take many many years. Easily a decade at least if they start now full fledged and everything goes right. Not an easy task.

x86 has around 1.5 to 2 billion users. And most of them won't consider an alternative until there's a compelling reason and/or forced to do so.

What if there is an emulator as good as Rosetta?

You don't really need native apps for smooth operation, as the emulation layer will do an excellent job.

 

[SiliconFly]

What if there is an emulator as good as Rosetta?

You don't really need native apps for smooth operation, as the emulation layer will do an excellent job.

I still remember the imaginary Windows Subsystem for Android (WSA). Imaginary cos it's useless in real life. Nice on paper, but sucks in reality. Very recently, I tried running some apps in Win10 in a VM on a Win11 system (running on excellent hardware) for some official stuff. Hell no. Emulation has limits. I love VMs, but native is cool. And like jim keller said, ISA is irrelevant. So, i'd say, we should pick the ISA we like and stay native for best results.

 

[Joe NYC]

But the LP cores for client will feature only in Zen 6

I am not sure what you mean by that.

AMD's Zen 4c are quite low power cores right now (compared to other x86 server cores).

Zen 6 cores will be 3rd generation of this approach for server. It seems that these cores may become the standard cores in Zen 6 generation, unlike Zen 4 and Zen 5, where the high performance cores are the "standard" cores.

But again, these are for server, not client.

. And it'll be sad if Zen 6 client still doesn't support NOC.

As far as Zen 6 client, the MLID leak had pretty much nothing on Zen 6 client. And the MLID leak is the only information that is floating around about Zen 6. So nothing is public about what is and what is not in Zen 6 client.

 

[FlameTail]

And what comes after Zen 6? MLID think Zen6 will be the last of the Zen architecture.

 

[Tigerick]

And what comes after Zen 6? MLID think Zen6 will be the last of the Zen architecture.

Next will be competitor to Intel's Nova Lake, whatever codename AMD will be given. Also the successor of X Elite and Apple's M5 series are both targeting for 2026 release

 

[igor_kavinski]

And who is this architect you speak of?

I'm hoping there is at least ONE restless CPU designer there. Otherwise, they are all lazy slobs content with their fat paychecks and that's exactly why Intel fell behind.

 

[SiliconFly]

I am not sure what you mean by that.

AMD's Zen 4c are quite low power cores right now (compared to other x86 server cores).

He says Zen 6 will have the usual Zen C cores and also newer and more efficient LP cores (like the LP E cores in MTL).

 

[Joe NYC]

He says Zen 6 will have the usual Zen C cores and also newer and more efficient LP cores (like the LP E cores in MTL).

He said that about what comes after Zen 6 (Venice), which may not another Zen generation but a set of new architectures. That's quite far away. If we assume 18 month cadence, we would have:
Zen 5: Mid 2024
Zen 6: End 2025
Next: Mid 2027

That's quite far away, unless some of these new architectures are developed in parallel.

In the meantime it would be a mistake to underestimate what the current, lower power Zen 4c, Zen 5c and Zen 6(c) can do. Here is a review of Zen 4c

https://twitter.com/x/status/1729929327582810256

 

[DrMrLordX]

Down with x86!
Long live ARM!

Down with ARM!
Long live RISC-V!

Sir, this is Wendy's.

 

[FlameTail]

Also the successor of X Elite are both targeting for 2026 release

WHAT IS THE SOURCE

 

[igor_kavinski]

Here is a review of Zen 4c

https://twitter.com/x/status/1729929327582810256

Just utter humiliation for Intel, except for a few benchmarks. And in power consumption, simply no contest. To offset the economics of running these power hungry Xeons, Intel would have to offer to pay the data center power bills to sway customers to their side. I've said this before. Intel should just bloody give up and license Epyc from AMD and have them fabbed in their foundries. Better for the world and better than embarrassing themselves thoroughly. Intel, stop making your junk server CPUs and send your CPU designers back to silicon design school.

 

[SiliconFly]

He said that about what comes after Zen 6 (Venice), which may not another Zen generation but a set of new architectures. That's quite far away. If we assume 18 month cadence, we would have:
Zen 5: Mid 2024
Zen 6: End 2025
Next: Mid 2027

That's quite far away, unless some of these new architectures are developed in parallel.

In the meantime it would be a mistake to underestimate what the current, lower power Zen 4c, Zen 5c and Zen 6(c) can do. Here is a review of Zen 4c

https://twitter.com/x/status/1729929327582810256

Forget AMD's C cores (Zen 4c, Zen 5c and Zen 6c). Intel's E cores by themselves are more efficient than AMD C corers & even ARM cores. Sierra Forrest with 288 cores is gonna kick the life out of competition.

And the newer Intel's LP E cores are expected to be even more efficient than the already efficient E cores. LNC is also being targeted towards efficiency. So is 20A. Intel's upcoming lineup is far better than what people perceive at the moment.

 

[TESKATLIPOKA]

Forget AMD's C cores (Zen 4c, Zen 5c and Zen 6c). Intel's E cores by themselves are more efficient than AMD C corers & even ARM cores. Sierra Forrest with 288 cores is gonna kick the life out of competition.

And the newer Intel's LP E cores are expected to be even more efficient than the already efficient E cores.

Proof?

 

[SiliconFly]

Proof?

It's linked.

 

[TESKATLIPOKA]

It's linked.

There is nothing like that there. It's showing unauthorized access If I want to check links in that text.

 

[Markfw]

Forget AMD's C cores (Zen 4c, Zen 5c and Zen 6c). Intel's E cores by themselves are more efficient than AMD C corers & even ARM cores. Sierra Forrest with 288 cores is gonna kick the life out of competition.

And the newer Intel's LP E cores are expected to be even more efficient than the already efficient E cores. LNC is also being targeted towards efficiency. So is 20A. Intel's upcoming lineup is far better than what people perceive at the moment.

An article from 2014 is supposed to have any value today ? I think not.

 

[Henry swagger]

Forget AMD's C cores (Zen 4c, Zen 5c and Zen 6c). Intel's E cores by themselves are more efficient than AMD C corers & even ARM cores. Sierra Forrest with 288 cores is gonna kick the life out of competition.

And the newer Intel's LP E cores are expected to be even more efficient than the already efficient E cores. LNC is also being targeted towards efficiency. So is 20A. Intel's upcoming lineup is far better than what people perceive at the moment.

Diamond rapids will be the game changer in data center.. stacked ai chiplets on cpu cores.. plus 16 wide decode if rentible units work as jim keller is planning 😄

 

[TESKATLIPOKA]

An article from 2014 is supposed to have any value today ? I think not.

So his proof that E-cores are more efficient is based on an article much older than e-cores. That’s just hilarious.

 

[FlameTail]

Also the successor of X Elite are both targeting for 2026 release

Source please?

 

[FlameTail]

Forget AMD's C cores (Zen 4c, Zen 5c and Zen 6c). Intel's E cores by themselves are more efficient than AMD C corers & even ARM cores. Sierra Forrest with 288 cores is gonna kick the life out of competition.

And the newer Intel's LP E cores are expected to be even more efficient than the already efficient E cores. LNC is also being targeted towards efficiency. So is 20A. Intel's upcoming lineup is far better than what people perceive at the moment.

What are you drinking?

I'd like to have some of it.