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

Tigerick · Aug 22, 2022

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

511 · Oct 15, 2024

https://twitter.com/x/status/1846236053541196120

Team x86 let's go

Dayman1225 · Oct 15, 2024

511 said:
https://twitter.com/x/status/1846236053541196120
Team x86 let's go

Interview of Lisa Su and Pat Gelsinger about it here

AMD, Intel Mobilize Major Tech Companies To Extend X86 Architecture

AMD's and Intel's CEOs explain why they have joined forces to create a blue-chip advisory board to shape the future of the ubiquitous x86 microprocessor architecture.

www.forbes.com

gdansk · Oct 15, 2024

It's good for both of them to agree on what the future of x86 is. They can't afford any split of the (shrinking) market at this point.

MoistOintment · Oct 15, 2024

Josh128 said:
What precisely do you use it for? Other than image generation which often requires a lot of manual tweaking once you obtain the output, web search summary is the only thing I use it for (because its default for most browsers now), and thats not particularly a must have feature to me.

Anything that benefits from matrix multiplication, really. Many workloads that are GPU accelerated can be performed on an NPU for better perf/watt.

Tensor cores, RTX cores, NPUs, Optical Flow accelerators - the future is *more* task specific accelerators, not less. General purpose CPU performance and GPU rasterization improvements are slowing, and progress will come from other sources.

Consumers already routinely use NPUs on their phones. From image processing and editing, separating foreground and background in images and video, search, text to speech, speech to text.

Intel even Demo'd some 3rd party software last week that runs on the NPU that allows physically handicap people to control games through face and head gestures in a Webcam.

Sure GPUs were previously used for many of these tasks, but NPUs are more specialized and can do the same job with less wattage.

NPUs absolutely aren't going anywhere. But their existence will fade into the background of consumer minds just like it already did on their phones.

As NPUs proliferate, developers will find use cases for them. We weren't going to see software that benefits from an NPU in widespread use *before* NPUs were in the market for a few years.

ShimmerBlade · Oct 15, 2024

https://twitter.com/x/status/1846245804966859195

Panther Lake smiles...

dullard · Oct 15, 2024

511 said:
https://twitter.com/x/status/1846236053541196120
Team x86 let's go

No one could have possibly predicted that.

dullard said:
Ehhh, all of the people who prefer one company over the other are misguided anyways. Intel and AMD are on the same x86 team. It is them vs ARM. Any infighting between AMD and Intel just hastens the death of both. This petty bickering is trying to win one portion of one battle while totally ignoring the war.

dullard · Oct 15, 2024

Josh128 said:
What precisely do you use it for? Other than image generation which often requires a lot of manual tweaking once you obtain the output, web search summary is the only thing I use it for (because its default for most browsers now), and thats not particularly a must have feature to me.

I tend to use it for writing a bunch of stuff. Whether my boss wants a summary of a big document, or if I need to write a carefully email delicately threading between a rock and a hard place. Or if I just don't know where to begin (condolence letters or similar).

I'd rather throw a dozen documents at AI and then proof read and correct AI's writing than do all the work myself, especially when it is for an unimportant check-the-box document that no one reads.

I tend to be extremely matter of fact (I'm probably further up on the Autism spectrum than most and also geeky). My writing is quite off-putting for some people. So, I frequently write something and then ask AI to make it friendlier. It seems to work quite well.

I normally like to edit my photos carefully and can do a pretty decent job at it. But, AI is almost as good and sometimes I'd rather spend my time doing other tasks. So, I'll ask AI to remove people or background problems from my images.

I've done just a little instant translation with AI. Helped when working with someone who speaks only Catalan and even the native Spanish person on my team couldn't understand.

I'd say it saves well over half of my time and gets better results.

Doug S · Oct 15, 2024

MoistOintment said:
As NPUs proliferate, developers will find use cases for them. We weren't going to see software that benefits from an NPU in widespread use *before* NPUs were in the market for a few years.

Yeah Microsoft sorta had to set some requirements, otherwise AMD and Intel would do what they do best, and market segment NPUs into a niche so it would take far longer before there were enough NPU equipped PCs for developers to bother.

Then they'd see all the new stuff AI can do coming on smartphones and Macs, with Windows being left behind.

adroc_thurston · Oct 15, 2024

Doug S said:
AI can do coming on smartphones

GEMM blobs dwelled on phones since late'17.
Where is the stuff?

dttprofessor · Oct 15, 2024

Npu & XMX are same ?

Saylick · Oct 15, 2024

ShimmerBlade said:
https://twitter.com/x/status/1846245804966859195
Panther Lake smiles...

Quick. Need someone to do a zoom and enhance on that die.

FlameTail · Oct 15, 2024

Saylick said:
Quick. Need someone to do a zoom and enhance on that die.

Is the resolution enough?

Also this die shot essentially confirms that Panther Lake won't have on-package memory like Lunar Lake.

https://twitter.com/x/status/1846246772366053471

I do wonder, will Panther Lake keep the multiple-PMIC design introduced by Lunar Lake? It's said that it improves efficiency.

dttprofessor · Oct 15, 2024

FlameTail said:
Is the resolution enough?

Also this die shot essentially confirms that Panther Lake won't have on-package memory like Lunar Lake.

https://twitter.com/x/status/1846246772366053471

I do wonder, will Panther Lake keep the multiple-PMIC design introduced by Lunar Lake? It's said that it improves efficiency.

No pmic

Hitman928 · Oct 15, 2024

511 said:
https://twitter.com/x/status/1846236053541196120
Team x86 let's go

Very long overdue, IMO.

cebri1 · Oct 15, 2024

FlameTail said:
Is the resolution enough?

Also this die shot essentially confirms that Panther Lake won't have on-package memory like Lunar Lake.

I do wonder, will Panther Lake keep the multiple-PMIC design introduced by Lunar Lake? It's said that it improves efficiency.

I believe that was already known. LNL may be a one off according to some people. It's like the LP-ecores from MTL, seems like Intel was looking into adhoc solution to be perf/watt competitive but once they get the manufacturing process up to speed and bring in the new archs there is really no need for custom solutions.

Although I do expect Lunar Lake succesor in 2026, with a two year release cadence to bring significant power savings gen-on-gen. Right now the N3B to 18A jump doesn't seem high enough to justify another product.

FlameTail · Oct 15, 2024

Surface Laptop 7 to be updated to have Lunar Lake?

Next-gen Surface Laptop with Intel Lunar Lake chip leaks as unauthorized sale appears online

An Intel Core Ultra 268V might make its way to the new Surface Laptop in the coming months.

www.windowscentral.com

Idf true, this means it's not an ARM/Qualcomm exclusive laptop, as many believed initially.

adroc_thurston · Oct 15, 2024

FlameTail said:
Idf true, this means it's not an ARM/Qualcomm exclusive laptop, as many believed initially.

Probably the commercial variant.

cebri1 said:
It's like the LP-ecores from MTL, seems like Intel was looking into adhoc solution to be perf/watt competitive but once they get the manufacturing process up to speed and bring in the new archs there is really no need for custom solutions.

No. The answer is cost.
It's just about cost. Need to get stuff cheaper and gooder.

DZero · Oct 15, 2024

I am thinking if we get the Alder Lake -N sucessor. Small cores of Lunar Lake are really impressive.

cannedlake240 · Oct 15, 2024

Saylick said:
Quick. Need someone to do a zoom and enhance on that die.

Pretty sure there was a diagram leak detailing tile dimensions on PTL. The main 18A compute tile will be ~115mm2 and N3E based 12Xe3 GPU over 50mm2

whoshere · Oct 15, 2024

Intel 18A could be working as intended as Pat Gelsinger has demonstrated the first ever sample of Panther Lake:

Saylick · Oct 15, 2024

cannedlake240 said:
Pretty sure there was a diagram leak detailing tile dimensions on PTL. The main 18A compute tile will be ~115mm2 and N3E based 12Xe3 GPU over 50mm2

Oh, you're right...

cebri1 · Oct 15, 2024

Saylick said:
Oh, you're right...
View attachment 109470

With the current DD, about 65% yield. Probably closer to 75-80% towards launch.

dullard · Oct 15, 2024

Saylick said:
Quick. Need someone to do a zoom and enhance on that die.

Is this good enough?

The Hardcard · Oct 15, 2024

dttprofessor said:
Npu & XMX are same ?

No, XMX is matrix math in the GPU.

DavidC1 · Oct 15, 2024

cebri1 said:
I believe that was already known. LNL may be a one off according to some people. It's like the LP-ecores from MTL, seems like Intel was looking into adhoc solution to be perf/watt competitive but once they get the manufacturing process up to speed and bring in the new archs there is really no need for custom solutions.

Manufacturing process won't result in a device with better battery life. That was true in the P3 days without advanced power management. Process basically only helps in load.

System-level engineering is required to get good battery life, and Pantherlake loses few key factors that Lunarlake has. It's probably going to be better than the horrible regressions of Alderlake-Raptorlake era but that's not saying much.

@adroc_thurston BoM is not the only important thing. If they lose marketshare again to Qualcomm in PTL-U then few $ difference in BoM won't matter, because they'll lose entire chip being in a design. A 10 year old can tell you 10% lower margin is preferable to zero revenue.

Same backward thinking as Otellini losing iPhone deal.

Although I do expect Lunar Lake succesor in 2026, with a two year release cadence to bring significant power savings gen-on-gen. Right now the N3B to 18A jump doesn't seem high enough to justify another product.

There are only few notable jumps. Pentium M, Haswell-Broadwell, Lunarlake. None of which are process related. Arrandale = regression, Alder/Raptor = regression.

DZero said:
I am thinking if we get the Alder Lake -N sucessor. Small cores of Lunar Lake are really impressive.

No Skymont based successor yet. The next one is Twin Lake, which is a refresh of Alderlake-N.

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

Senior member

Attachments

Senior member

Golden Member

Diamond Member

Member

Junior Member

Elite Member

Elite Member

Platinum Member

Diamond Member

Member

Diamond Member

Diamond Member

Member

Diamond Member

Senior member

Diamond Member

Diamond Member

Member

Member

Member

Diamond Member

Senior member

Elite Member

Senior member

Golden Member