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 · May 22, 2025

Linux changes for many of the lunar lake were not merged in the kernel version he tested so it was kind of a beta test initially.

dullard · May 23, 2025

Hulk said:
Nice find Igor.
I wonder how much of this is just due to bumping up power/heat though?

Linux just fixed an Intel power issue (the NPU was never put to sleep and kept using power even when idle): https://www.phoronix.com/news/Linux-6.15-Fix-ARL-U-H-s2idle

As for the desktop Arrow Lake, Phoronix found a 6% speed improvement with a 3% average power reduction. This came from Linux changes putting the software on the proper cores (see the first two graphs here):

Intel Core Ultra 9 285K "Arrow Lake" Performance On Linux Has Improved A Lot Since Launch Review - Phoronix

www.phoronix.com

Just saying that performance changes and power changes can very well be independent from each other. This is especially true when drivers or the OS are operating a CPU out of the optimum conditions.

Tigerick · May 24, 2025

Tigerick said:
Intel is claiming 20% performance uplift. 4+8+4 we have 16 cores on the same die (18A) compared to 14 main CPU cores

Last year, my source told me Intel was claiming 20% performance uplift compared to ArrowLake H with 6+8+2 cores. Now Intel is claiming similar performance with ARL-H. Well, depending on the final clock speed, I am actually expecting worse performance than ARL-H @ 5.4GHz max. Sure, iGPU will perform faster thanks to TSMC's N3E process with 12 Xe cores but not CPU fabbed by IFS's 18A...

Come on, how hard could 4+8+4 compute die be made? Turn out very hard for IFS, that's why IFS cancelled 20A process and completely forgotten about 6+8 exists in the roadmap...

511 · May 24, 2025

Tigerick said:
Last year, my source told me Intel was claiming 20% performance uplift compared to ArrowLake H with 6+8+2 cores. Now Intel is claiming similar performance with ARL-H. Well, depending on the final clock speed, I am actually expecting worse performance than ARL-H @ 5.4GHz max. Sure, iGPU will perform faster thanks to TSMC's N3E process with 12 Xe cores but not CPU fabbed by IFS's 18A...

20% MT? Intel claimed LNL efficiency with ARLH MT IDK what that means.

Tigerick said:
Come on, how hard could 4+8+4 compute die be made? Turn out very hard for IFS, that's why IFS cancelled 20A process and completely forgotten about 6+8 exists in the roadmap...

It's just all rumors coming from Taiwan. Also they have a money problem now a very big one when 20A was supposed to ramp. They didn't have money to ramp 🤣.

Tigerick · May 24, 2025

511 said:
20% MT? Intel claimed LNL efficiency with ARLH MT IDK what that means.

Oh, about that slide actually compared to PTL-H and PTL-U series seperately. Thanks to your X links about PTL-H SKUs; what is missing is PTL-U SKUs which are supposed to be released around the same timeframe.

Basically, PTL-U is a ultra-portable SoC with 4P + 4LPe (8 e cores are disabled) and 4XE cores (Intel 3-E). PTL-U are most likely replacing ARL-U but not LNL-MX. LNL-MX is more power efficiency and comes with 8 XE cores.

511 said:
It's just all rumors coming from Taiwan. Also they have a money problem now a very big one when 20A was supposed to ramp.

511 · May 24, 2025

Tigerick said:
Oh, about that slide actually compared to PTL-H and PTL-U series seperately. Thanks to your X links about PTL-H SKUs; what is missing is PTL-U SKUs which are supposed to be released around the same timeframe.

I think the marketing is a mess after the 35% cut.

Tigerick said:
Basically, PTL-U is a ultra-portable SoC with 4P + 4LPe (8 e cores are disabled) and 4XE cores (Intel 3-E). PTL-U are most likely replacing ARL-U but not LNL-MX. LNL-MX is more power efficiency and comes with 8 XE cores.

I am not really sure about the CPU part being more efficient than LNL vs 4+0+4 PTL-U the GPU can be but not other parts cause everything is brand new and upgraded in PTL.

511 · May 28, 2025

What is Intel even cooking with these 😂

igor_kavinski · May 28, 2025

Frankly, looks hideous.

511 · May 28, 2025

Yeah but this is going to pack UCC Diamond Rapids

AcrosTinus · May 29, 2025

511 said:
What is Intel even cooking with these 😂View attachment 124478

Big lane and core increases with required power increase of 700W+ are on the horizon with this.

511 · May 29, 2025

AcrosTinus said:
Big lane and core increases with required power increase of 700W+ are on the horizon with this.

Well I read a serve the home article that says there are 16 DIMM Slots on each side that would mean this is a freaking 16 Channel platform

Auras Shows Next-Gen Intel Oak Stream Cold Plates at Computex 2025

At Computex 2025, we saw Intel Oak Stream liquid cooling cold plates for next-generation Intel Xeon processors

www.servethehome.com

AcrosTinus · May 29, 2025

511 said:
Well I read a serve the home article that says there are 16 DIMM Slots on each side that would mean this is a freaking 16 Channel platform

Auras Shows Next-Gen Intel Oak Stream Cold Plates at Computex 2025

At Computex 2025, we saw Intel Oak Stream liquid cooling cold plates for next-generation Intel Xeon processors

www.servethehome.com

Wouldn't that throughput be needed for a single socket scale up, like more cores, lanes and even higher PCIe generation ?

511 · May 29, 2025

AcrosTinus said:
Wouldn't that throughput be needed for a single socket scale up, like more cores, lanes and even higher PCIe generation ?

Yeah it will be there Is also DDR5 12800 MT/s MRDIMMS for DMR delivering 1.63 TB/s Bandwidth!. That's more than 4090 and just less than that of 1.79 TB/s for 5090.

MS_AT · May 29, 2025

511 said:
Yeah it will be there Is also DDR5 12800 MT/s MRDIMMS for DMR delivering 1.63 TB/s Bandwidth!. That's more than 4090 and just less than that of 1.79 TB/s for 5090.

Hopefully most of it will be usable, compared to the Sapphire Rapid HBM fiasco where out of theoretical 1.60TB/s only 700-600GB/s is usable by the CPU. Of course with DDR5 there is also a difference between practical and theoretical BW, but usually it's 70-80%

https://www.ixpug.org/images/docs/ISC23/McCalpin_SPR_BW_limits_2023-05-24_final.pdf

yuri69 · May 29, 2025

511 said:
Well I read a serve the home article that says there are 16 DIMM Slots on each side that would mean this is a freaking 16 Channel platform

The 2026 AMD Venice is reportedly split between a highend SP7 socket with 16ch and SP8 socket with a lower channel count. So yea, it's expensive but nothing extraordinary for 2026-2027 products.

511 · May 29, 2025

MS_AT said:
Hopefully most of it will be usable, compared to the Sapphire Rapid HBM fiasco where out of theoretical 1.60TB/s only 700-600GB/s is usable by the CPU. Of course with DDR5 there is also a difference between practical and theoretical BW, but usually it's 70-80%

Look what i found looks like something is bottle necking GNR in 2S config i wonder what are the values for 1S.

Granite Rapids MRDIMM Vs. DDR5 Benchmarks Performance - OpenBenchmarking.org

openbenchmarking.org

Zen 6 is 16 channel with DDR5 8000 IIRC so approx 1 TB/s.

Fjodor2001 · May 30, 2025

Arrow Lake-S refresh coming after all:

Intel Arrow Lake-S Refresh CPUs Confirmed: Coming To LGA 1851 & 800-Series Motherboard Platforms

Intel's Arrow Lake-S Refresh "Core Ultra Series 2" CPUs are now confirmed and will be supported by existing LGA 1851 socketed motherboards.

wccftech.com

AcrosTinus · May 30, 2025

Fjodor2001 said:
Arrow Lake-S refresh coming after all:

Intel Arrow Lake-S Refresh CPUs Confirmed: Coming To LGA 1851 & 800-Series Motherboard Platforms

Intel's Arrow Lake-S Refresh "Core Ultra Series 2" CPUs are now confirmed and will be supported by existing LGA 1851 socketed motherboards.

wccftech.com

They surprised me in a positive and negative sense with Raptor Cove, the gaming improvements from the 12900K were crazy, even a 13600K could surpass it. Can they pull that again, this time without degradation ?

DavidC1 · May 30, 2025

Not sure why you are expecting magic with the Refresh? It's likely just Arrowlake but 100-200MHz higher clocks.

Fjodor2001 · May 31, 2025

DavidC1 said:
Not sure why you are expecting magic with the Refresh? It's likely just Arrowlake but 100-200MHz higher clocks.

And a faster NPU according to the article.

511 · May 31, 2025

Fjodor2001 said:
And a faster NPU according to the article.

Oh the fast NPU I can't believe it finally I can get a token/s more oh Lord such a massive improvement.

Fjodor2001 · May 31, 2025

511 said:
Oh the fast NPU I can't believe it finally I can get a token/s more oh Lord such a massive improvement.

I didn’t say it was a massive improvement, just mentioned the info from the article.

511 · May 31, 2025

Fjodor2001 said:
I didn’t say it was a massive improvement, just mentioned the info from the article.

i was being sarcastic over the improvement to NPU as intel choose to focus on this.

Fjodor2001 · May 31, 2025

511 said:
i was being sarcastic over the improvement to NPU as intel choose to focus on this.

Likely it’s to pass some Microsoft Copilot AI TOPS threshold requirement so they can put a necessary sticker for that on the PC box. Might affect sales if the average Joe is told to look for that when buying a new PC.

AcrosTinus · May 31, 2025

DavidC1 said:
Not sure why you are expecting magic with the Refresh? It's likely just Arrowlake but 100-200MHz higher clocks.

I am still hoping for some tweaks to the cache like they did with Raptor Lake, It might be cope but 12900K -> 13900K is the gain I am hoping for not 13900K -> 14900K, they have two options to surprise me with.

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

Senior member

Attachments

Platinum Member

Elite Member

Senior member

Platinum Member

Senior member

Platinum Member

Platinum Member

Lifer

Platinum Member

Member

Platinum Member

Member

Platinum Member

Senior member

Senior member

Platinum Member

Diamond Member

Member

Golden Member

Diamond Member

Platinum Member

Diamond Member

Platinum Member

Diamond Member

Member