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.)

 

[lightisgood]

https://videocardz.com/newz/first-g...lake-soc-with-3ghz-arc-a5-gpu-has-been-demoed. 3ghz on gpu on a handheld tsmc 5nm is very efficient.. valve needs a intel powered steamdeck for the steamdeck 2 💻😃

MTL should beat Phoenix. MLID was correct...

 

[mikk]

https://videocardz.com/newz/first-g...lake-soc-with-3ghz-arc-a5-gpu-has-been-demoed. 3ghz on gpu on a handheld tsmc 5nm is very efficient.. valve needs a intel powered steamdeck for the steamdeck 2 💻😃

3 Ghz is most likely a misread. It doesn't even change when he changes the power level on the fly unlike the CPU speed. Software isn't final, such things can happen on a new platform. I wonder if Arc 5 is the full 8 Xe variant or 7 Xe variant.

 

[DrMrLordX]

Given Intel's classic aversion to low console margins, I doubt they'll be making any deals with Valve. Plus Meteor Lake parts are going to be expensive for awhile.

 

[igor_kavinski]

Plus Meteor Lake parts are going to be expensive for awhile.

Remains to be seen. I have a bad feeling about MTL performance, other than its tGPU.

 

[DrMrLordX]

Remains to be seen. I have a bad feeling about MTL performance, other than its tGPU.

The performance won't dictate price. Not for awhile anyway.

 

[lightisgood]

Given Intel's classic aversion to low console margins, I doubt they'll be making any deals with Valve.

Intel dosen't attach importance to UMPC market.
However, it dosen't mean that Intel is very particular about customer.

 

[Geddagod]

MTL should beat Phoenix. MLID was correct...

I wouldn't be too surprised if the iGPU synthetics look pretty good... only to be a lackluster when looking at perf in games themselves. From what I hear, that's the general trend with Intel's iGPUs recently
Also, MTL beating phoenix, overall, as a product? Doubt. Guessing it's going to be close though.

 

[S'renne]

I wouldn't be too surprised if the iGPU synthetics look pretty good... only to be a lackluster when looking at perf in games themselves. From what I hear, that's the general trend with Intel's iGPUs recently
Also, MTL beating phoenix, overall, as a product? Doubt. Guessing it's going to be close though.

Hope its close enough though, but even if it isn't, it still offers similar features I think, and still has more availability than Phoenix later down the line

 

[Geddagod]

Hope its close enough though, but even if it isn't, it still offers similar features I think, and still has more availability than Phoenix later down the line

Only reason it might end up having more availability than Phoenix wouldn't be because Intel is pumping out MTL, but because AMD just sucks at mobile volume. I think the existence of RPL and RPL-R mobile and MTL launching so late is pretty indicative of that.

 

[S'renne]

Only reason it might end up having more availability than Phoenix wouldn't be because Intel is pumping out MTL, but because AMD just sucks at mobile volume. I think the existence of RPL and RPL-R mobile and MTL launching so late is pretty indicative of that.

I'm just waiting on MTL efficiency benchmarks anyways since my real interest is Arrow Lake, I want a desktop replacement laptop with ARL lol

 

[mikk]

I wouldn't be too surprised if the iGPU synthetics look pretty good... only to be a lackluster when looking at perf in games themselves. From what I hear, that's the general trend with Intel's iGPUs recently
Also, MTL beating phoenix, overall, as a product? Doubt. Guessing it's going to be close though.

With a 2x improvement over Iris Xe they can easily match Phoenix. When it comes to 3dmark I would use timespy graphics, I think it needs roughly 3500 points to match Phoenix. This is based on A380 vs RX6400. A380 scores 15-20% higher whereas in real world they are evenly matched. The fastest Phoenix 780M devices score 2800-2900 points. 15-20% higher than this gives roughly 3500 points. I think this is doable for MTL-H tGPU at 2.2 Ghz.

 

[DrMrLordX]

Intel dosen't attach importance to UMPC market.
However, it dosen't mean that Intel is very particular about customer.

Yes, but please keep in mind that Valve DOES attach importance to BoM.

 

[TESKATLIPOKA]

With a 2x improvement over Iris Xe they can easily match Phoenix. When it comes to 3dmark I would use timespy graphics, I think it needs roughly 3500 points to match Phoenix. This is based on A380 vs RX6400. A380 scores 15-20% higher whereas in real world they are evenly matched. The fastest Phoenix 780M devices score 2800-2900 points. 15-20% higher than this gives roughly 3500 points. I think this is doable for MTL-H tGPU at 2.2 Ghz.

It probably is doable at 2.2GHz, but what I want to know is the power limit for that frequency.
If 35-45W is enough for that then great, If more is needed then not great.

 

[mikk]

It probably is doable at 2.2GHz, but what I want to know is the power limit for that frequency.
If 35-45W is enough for that then great, If more is needed then not great.

Phoenix requires more than 50W to reach 2800-2900 points, the fastest one use 70-80W.

 

[Geddagod]

https://www.tomshardware.com/news/intel-core-ultra-7-155h-underwhelms-in-passmark-database-appearance

MT score is... questionable, though we don't know what TDP is used
ST score of >5%. Not good, but not bad either. Intel doesn't need a massive ST bump to compete with Phoenix, so it's not too disappointing.

Oh also, talk on Twitter from Raichu +Bionic is that :
ARL-H GPU tile uses N4P, and it has some improvements and is not a complete refresh of the MTL-H iGPU die (doubled L2 and XMX). Though it's not BMG.
ARL-S reuses the 64 EU GPU tile of MTL though

 

[H433x0n]

I wouldn't be too surprised if the iGPU synthetics look pretty good... only to be a lackluster when looking at perf in games themselves. From what I hear, that's the general trend with Intel's iGPUs recently
Also, MTL beating phoenix, overall, as a product? Doubt. Guessing it's going to be close though.

What’s considered beating Phoenix overall? You don’t think it’ll manage a 15-20% boost to performance at isopower sub 55W?

Edit: That link you posted has an odd MT score. Crestmont has ~5% IPC increase and most MTL SKUs will have similar (or greater) e-core clocks over RPL. This benchmark had to be power or thermal limited.

 

[lightisgood]

Probably, low power limit generates low MT score.
I'm very optimistic about MTL MT performance.

By the way, I have a great interest in 5.6% ST score boost rather than low MT score.
Intel said that Redwood Cove has Rapter Cove's IPC on the whole, didn't it?
I don't think 5.6% ST boost is small difference....
Intel provided sloppy explanation about Redwood Cove, RIGHT?
I'd like to see other ST benchmark cautiously.

 

[H433x0n]

Probably, low power limit generates low MT score.
I'm very optimistic about MTL MT performance.

By the way, I have a great interest in 5.6% ST score boost rather than low MT score.
Intel said that Redwood Cove has Rapter Cove's IPC on the whole, didn't it?
I don't think 5.6% ST boost is small difference....
Intel provided sloppy explanation about Redwood Cove, RIGHT?
I'd like to see other ST benchmark cautiously.

5.6% ST increase seems too good to be true tbh. I doubt that ends up being a representative result.

 

[Geddagod]

What’s considered beating Phoenix overall? You don’t think it’ll manage a 10-12% boost to performance at isopower sub 55W?

It's closer to 15-20%, which is cutting it really close. I think it will tie or maybe even slightly lose to Phoenix tbh. Also, in single core perf/watt (for RWC), I believe it will be worse there as well.
In iGPU, it appears to be better
In battery life, I mean Intel claims 2x, but that's highly dependent on how they test battery life. I expect a massive improvement, but there's a lot of ground to cover too.
And in ST peak performance, I expect them to be roughly tied.

This isn't an overall win over Phoenix, since they have some areas where they might lose (albeit slightly) or be tied in. What's even worse is how much later it's coming compared to Phoenix as well, and how much more expensive it looks to be too. At best one could say this is a slightly buffed Phoenix, depending on what metric you are comparing it for.

 

[S'renne]

It's closer to 15-20%, which is cutting it really close. I think it will tie or maybe even slightly lose to Phoenix tbh. Also, in single core perf/watt (for RWC), I believe it will be worse there as well.
In iGPU, it appears to be better
In battery life, I mean Intel claims 2x, but that's highly dependent on how they test battery life. I expect a massive improvement, but there's a lot of ground to cover too.
And in ST peak performance, I expect them to be roughly tied.

This isn't an overall win over Phoenix, since they have some areas where they might lose (albeit slightly) or be tied in. What's even worse is how much later it's coming compared to Phoenix as well, and how much more expensive it looks to be too. At best one could say this is a slightly buffed Phoenix, depending on what metric you are comparing it for.

Btw doesn't Meteor Lake's NPU has better software support than AMD's Xilinx NPU from their demo? They at least also have better software support compared to AMD in that area

 

[Geddagod]

Btw doesn't Meteor Lake's NPU has better software support than AMD's Xilinx NPU from their demo? They at least also have better software support compared to AMD in that area

Tbh completely forgot about the entire AI element of both Intel and AMD mobile chips. No idea ¯\_(ツ)_/¯

 

[TESKATLIPOKA]

Phoenix requires more than 50W to reach 2800-2900 points, the fastest one use 70-80W.

You don't need that much power.

Ryzen Z1 Extreme => PL4:53W, PL2:43W, PL1:30W
Time Spy Graphics: 2854 points
PL4 duration is ~10s
PL2 duration is ~450s

TimeSpy will finish within PL2 duration, so 43W highest for Phoenix.

Comparable performance at ~45W would be great for Meteor Lake.

 

[H433x0n]

It's closer to 15-20%, which is cutting it really close. I think it will tie or maybe even slightly lose to Phoenix tbh. Also, in single core perf/watt (for RWC), I believe it will be worse there as well.
In iGPU, it appears to be better
In battery life, I mean Intel claims 2x, but that's highly dependent on how they test battery life. I expect a massive improvement, but there's a lot of ground to cover too.
And in ST peak performance, I expect them to be roughly tied.

This isn't an overall win over Phoenix, since they have some areas where they might lose (albeit slightly) or be tied in. What's even worse is how much later it's coming compared to Phoenix as well, and how much more expensive it looks to be too. At best one could say this is a slightly buffed Phoenix, depending on what metric you are comparing it for.

I'd take that outcome tbh. Phoenix is on N4P, arguably the best TSMC has to offer at the moment. It's also going against Zen 4, probably the best overall core design in the past 10 years. If it matches Phoenix efficiency (+ or - 5%) I would say that's the best position Intel has had in a while especially since they can actually provide volume and support.

 

[S'renne]

Tbh completely forgot about the entire AI element of both Intel and AMD mobile chips. No idea ¯\_(ツ)_/¯

Part of the main selling point of both Meteor Lake and Phoenix Point is the AI element I'm pretty sure(even though most people don't care instead)

 

[mikk]

You don't need that much power.

Ryzen Z1 Extreme => PL4:53W, PL2:43W, PL1:30W
Time Spy Graphics: 2854 points
PL4 duration is ~10s
PL2 duration is ~450s

TimeSpy will finish within PL2 duration, so 43W highest for Phoenix.

Comparable performance at ~45W would be great for Meteor Lake.

This is a special binned and tuned model for handheld only. In a regular notebook most OEMs go for performance to gain just a few percent which require 20-40W more power, although I would expect a handheld OEM even with a regular SKU won't be so crazy.

Oh also, talk on Twitter from Raichu +Bionic is that :
ARL-H GPU tile uses N4P, and it has some improvements and is not a complete refresh of the MTL-H iGPU die (doubled L2 and XMX). Though it's not BMG.
ARL-S reuses the 64 EU GPU tile of MTL though

He also said the GT3 version had been cancelled. This was expected to me. GT2 with unchanged 128EUs is disappointing if true. This comes with a limited increase in GPU performance in the next generation. ADM is also unlikely without GT3. It's a poor execution that they couldn't switch to Xe2, Lunar Lake gets it in a similar launch timeframe.