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 (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.)

 

[Antey]

If the media engine and display engine is in the SoC tile, and Intel is placing a new 'low power island' in the SoC for low power usage. Why not place a very small and simplier GPU cores, cache, etc with a ver low power consumption design (like adreno/mali cores) in the SoC for a lower power usage in the same way? i guess it would need lots of other things like a an unified memory controller, not an easy wish.

I love that idea of having a very power efficient SoC tile and whenever you need performance Graphics/CPU... tiles wake up. It's brilliant.

In the future, wouldnt it be incredible to have the possibilty of SOCs with low power islands that are good enough for browsing, light programming, light microsoft office, etc, fanless notebooks/PCs, fully noiseless, and that extra possibility that a compute/gpu tile gives, you can have a 4090 in a GPU tile and 32+ P-Cores/E-cores in a CPU tile sleeping, just waiting for when you need it, but at the same time consuming less than <15W (when CPU/GPU tiles are in sleep mode)? and just to think that the Intel 12900k consumes 55W idle...

 

[trivik12]

just to think that the Intel 12900k consumes 55W idle...

Are you sure if you dont have anything faulty.

Intel Core i9-12900K gaming performance review -- New highs, higher lows

See what the Intel 12900K can do with its eight P-cores and eight E-cores in our gaming performance review, and if it's worth the investment.

www.pcinvasion.com

Before we get to the game data, here is what to expect out of the 12900K at idle on Windows 11. The processor averaged roughly 27W with all power consumption preferences set to maximum. This idle power consumption can be reduced to a mere 4.72W average with the default Windows Power Saver plan. Under this power plan, the 12900K retained a reasonable degree of system responsiveness and averaged 8.77W. Even Adobe Photoshop was generally responsive and usable for basic photo editing.

 

[Antey]

i was using techpowerup data, but it's a whole system power consumption. Well, i guess 27W idle is not that good either. And that power plan hurts performance too much even if it ''retained a reasonable degree of system responsiveness''... meteor lake with a Low power island should be way more power efficient... <10W idle or even <5W idle and <1W with that kind of power plan

 

[DrMrLordX]

Honestly, if I were Intel I would have avoided the Raptor Lake Refresh frequency jump. Don’t even release a 14900k. Increase the core counts of lower tier parts, sure. That should keep OEMs happy.

It's an attempt to retake performance crowns for the sake of publicity. Nothing more. The 14900k is probably not going to be a good product except for reviewers that ask the wrong questions.

 

[H433x0n]

It's an attempt to retake performance crowns for the sake of publicity. Nothing more. The 14900k is probably not going to be a good product except for reviewers that ask the wrong questions.

It's going to be horrible publicity. It’s easy to predict how it’s going to go. I predict that HUB’s review in particular will show zero performance improvement over 13900K and talk about power consumption for 30 minutes.

Assuming motherboard vendors continue to set default with “MCE” active and no power limits it’s going to get miserable reviews.

 

[Antey]

Honestly, whats the difference between a 14900K and a 13900KS?

At least the 10900k added 2 cores over the 9900k and 200mhz, 11900k at least a different microarch (even if the final result wasnt that great or even worse)... but this one?

 

[H433x0n]

Honestly, whats the difference between a 14900K and a 13900KS?

At least the 10900k added 2 cores over the 9900k and 200mhz, 11900k atleast changed archs... but this one?

It's the 13900K with a tweaked Intel 7 process. The final result is +200mhz clocks at isopower. The 13900KS was +100mhz all core and +200mhz boost clocks but with additional power consumption. It's not a revolutionary product, it's a stopgap solution. It is unambiguously better than the 13900KS though.

13900K:
base clocks- 3.0ghz (125W)
boost clock - 5.8ghz
all core clock - 5.5ghz (=253W)

13900KS:
base clocks- 3.2ghz (150W)
boost clock - 6.0ghz
all core clock - 5.6ghz (>253W)

14900K:
base clocks- 3.2ghz (125W)
boost clock - 6.0ghz
all core clock - 5.7ghz (=253W)

 

[Geddagod]

It's the 13900K with a tweaked Intel 7 process. The final result is +200mhz clocks at isopower. The 13900KS was +100mhz all core and +200mhz boost clocks but with additional power consumption. It's not a revolutionary product, it's a stopgap solution. It is unambiguously better than the 13900KS though.

13900K:
base clocks- 3.0ghz (125W)
boost clock - 5.8ghz
all core clock - 5.5ghz (=253W)

13900KS:
base clocks- 3.2ghz (150W)
boost clock - 5.8ghz
all core clock - 5.6ghz (>253W)

14900K:
base clocks- 3.2ghz (125W)
boost clock - 6.0ghz
all core clock - 5.7ghz (=253W)

13900ks goes to 6 GHz

 

[H433x0n]

13900ks goes to 6 GHz

You're right, fixed it.

 

[Antey]

 

[H433x0n]

View attachment 86489

Pretty much. It's safe to say that 2023 is going to be a snooze fest for tech enthusiasts.

 

[coercitiv]

i was using techpowerup data, but it's a whole system power consumption. Well, i guess 27W idle is not that good either. And that power plan hurts performance too much even if it ''retained a reasonable degree of system responsiveness''... meteor lake with a Low power island should be way more power efficient... <10W idle or even <5W idle and <1W with that kind of power plan

Desktop Alder Lake uses 3-5W in idle, and there is no perceivable system responsiveness penalty to pay. Figures like 27W are only attainable with sleep states disabled, which is not stock config. Windows power plan can stay on Balanced or High Performance.

 

[Hulk]

It's the 13900K with a tweaked Intel 7 process. The final result is +200mhz clocks at isopower. The 13900KS was +100mhz all core and +200mhz boost clocks but with additional power consumption. It's not a revolutionary product, it's a stopgap solution. It is unambiguously better than the 13900KS though.

13900K:
base clocks- 3.0ghz (125W)
boost clock - 5.8ghz
all core clock - 5.5ghz (=253W)

13900KS:
base clocks- 3.2ghz (150W)
boost clock - 6.0ghz
all core clock - 5.6ghz (>253W)

14900K:
base clocks- 3.2ghz (125W)
boost clock - 6.0ghz
all core clock - 5.7ghz (=253W)

Where did you find the 5.7GHz all core clock spec for the 14900K? I have been curious about that.

 

[H433x0n]

Where did you find the 5.7GHz all core clock spec for the 14900K? I have been curious about that.

It’s from the geekbench logs. The maximum frequency listed for the multi core benchmarks usually say something like “5686mhz”, whereas when I run the test on my 13900K that same log says “5489mhz”.

That coupled with the base clock being +200mhz higher makes me feel pretty confident in the 5.7ghz all core clock prediction.

 

[Hulk]

It’s from the geekbench logs. The maximum frequency listed for the multi core benchmarks usually say something like “5686mhz”, whereas when I run the test on my 13900K that same log says “5489mhz”.

That coupled with the base clock being +200mhz higher makes me feel pretty confident in the 5.7ghz all core clock prediction.

That makes sense. I like it. Thanks.

 

[Geddagod]

Credit: Harukaze
Looks like ARL has yet to hit ES1.

 

[DrMrLordX]

Looks like ARL has yet to hit ES1.

Not really a problem given the launch window.

 

[mikk]

View attachment 86507
Credit: Harukaze
Looks like ARL has yet to hit ES1.

You don't know how old this is. Given that Lunar Lake is up and running AI tasks I would assume ARL-S ES1 is up and running as well.

 

[Hulk]

Are these test chips running on the node that is planned for production?
What I mean specifically is are Arrow Lake test parts built on 20A and are Lunar Lake test parts on 18A?
My follow up question would be how does Intel make it cost effective to sample the nodes for these parts in small quantities of for testing. AFAIK the machines cost 150million, can they be used for short runs? Can they be quickly changed from one node to another?
Is the production of the samples completely different production-wise from an actual production run of parts for sale?
I know that you can't easily change nodes as an architecture is designed for a particular node. I'm curious about how this works?

 

[Dayman1225]

Are these test chips running on the node that is planned for production?
What I mean specifically is are Arrow Lake test parts built on 20A and are Lunar Lake test parts on 18A?
My follow up question would be how does Intel make it cost effective to sample the nodes for these parts in small quantities of for testing. AFAIK the machines cost 150million, can they be used for short runs? Can they be quickly changed from one node to another?
Is the production of the samples completely different production-wise from an actual production run of parts for sale?
I know that you can't easily change nodes as an architecture is designed for a particular node. I'm curious about how this works?

Can’t answer most of the question but Lunar Lake is N3B not 18A

 

[igor_kavinski]

Can’t answer most of the question but Lunar Lake is N3B not 18A

Your source? This says it's 18A: https://videocardz.com/newz/intel-c...-cove-p-core-and-skymont-e-core-architectures

 

[Ajay]

Are these test chips running on the node that is planned for production?
What I mean specifically is are Arrow Lake test parts built on 20A and are Lunar Lake test parts on 18A?
My follow up question would be how does Intel make it cost effective to sample the nodes for these parts in small quantities of for testing. AFAIK the machines cost 150million, can they be used for short runs? Can they be quickly changed from one node to another?
Is the production of the samples completely different production-wise from an actual production run of parts for sale?
I know that you can't easily change nodes as an architecture is designed for a particular node. I'm curious about how this works?

Intel has at least one R&D lab in Oregon where they run smaller batches of wafers. Stands to reason that they have EUV powered lithography machines there. I haven’t read much about Intel R$D fabs for years, so I don’t know the particulars.

 

[igor_kavinski]

Intel has at least one R&D lab in Oregon where they run smaller batches of wafers. Stands to reason that they have EUV powered lithography machines there.

That's quite possibly right.

How Many Manufacturing Fabs Does Intel Have?

Learn more about the number and worldwid

www.intel.com

Don't think they would have an expensive machine delivered to anywhere other than their single testing site in the US:

 

[H433x0n]

Are these test chips running on the node that is planned for production?
What I mean specifically is are Arrow Lake test parts built on 20A and are Lunar Lake test parts on 18A?
My follow up question would be how does Intel make it cost effective to sample the nodes for these parts in small quantities of for testing. AFAIK the machines cost 150million, can they be used for short runs? Can they be quickly changed from one node to another?
Is the production of the samples completely different production-wise from an actual production run of parts for sale?
I know that you can't easily change nodes as an architecture is designed for a particular node. I'm curious about how this works?

New process tech is developed in Oregon.

It begins with research stage where the characteristics of the node are developed (number of layers, chemistry, etc.). This transitions to Initial Development where they try to get the process to yield. After initial development they attempt volume production and if that works it’s ready to be exported to another Intel fab to enter HVM (I think this is what they mean by manufacturing ready).

So a product like ARL on a new process could most likely have ES chips fabbed at Oregon.

 

[jpiniero]

Your source? This says it's 18A: https://videocardz.com/newz/intel-c...-cove-p-core-and-skymont-e-core-architectures

Why not both?