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

[Tigerick]

Wildcat Lake (WCL) Preliminary Specs

Intel Wildcat Lake (WCL) is upcoming mobile SoC replacing ADL-N. WCL consists of 2 tiles: compute tile and PCD tile. It is true single die consists of CPU, GPU and NPU that is fabbed by 18-A process. Last time I checked, PCD tile is fabbed by TSMC N6 process. They are connected through UCIe, not D2D; a first from Intel. Expecting launching in Q2/Computex 2026. In case people don't remember AlderLake-N, I have created a table below to compare the detail specs of ADL-N and WCL. Just for fun, I am throwing LNL and upcoming Mediatek D9500 SoC.

		Intel Alder Lake - N	Intel Wildcat Lake	Intel Lunar Lake		Mediatek D9500
Launch Date		Q1-2023	Q2-2026 ?	Q3-2024		Q3-2025
Model		Intel N300	?	Core Ultra 7 268V		Dimensity 9500 5G
Dies		2	2	2		1
Node		Intel 7 + ?	Intel 18-A + TSMC N6	TSMC N3B + N6		TSMC N3P
CPU		8 E-cores	2 P-core + 4 LP E-cores	4 P-core + 4 LP E-cores		C1 1+3+4
Threads		8	6	8		8
Max Clock		3.8 GHz	?	5 GHz
L3 Cache		6 MB	?	12 MB
TDP		7 W	Fanless ?	17 W		Fanless
Memory		64-bit LPDDR5-4800	64-bit LPDDR5-6800 ?	128-bit LPDDR5X-8533		64-bit LPDDR5X-10667
Size		16 GB	?	32 GB		24 GB ?
Bandwidth			~ 55 GB/s	136 GB/s		85.6 GB/s
GPU		UHD Graphics		Arc 140V		G1 Ultra
EU / Xe		32 EU	2 Xe	8 Xe		12
Max Clock		1.25 GHz		2 GHz
NPU		NA	18 TOPS	48 TOPS		100 TOPS ?

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.

 

[DavidC1]

The only major regressions I see are the lack of on-package memory, less use of PMICs, and possibly power loss with inter-die communication, but the biggest part of efficiency, the actual architecture and layout of the cores, are still applicable to panther lake AFAIK. I can definitely see like a 20-30% efficiency regression especially in very specific light workloads, like maybe in a scrolling webpage for 24 hours test it'd fall behind, but overall I'd say that PTL-U will be close in efficiency to lunar lake.

I said on package memory(not enough alone but LNL has PHY optimizations) alone is enough for 0.5-1W reduction which is substantial.

20-30% reduction makes it closer to Meteorlake than Lunarlake, and that's a huge amount. You are going from 15 hours to 10-12 hours.

I told you the inter-die communication effects aren't necessarily about pJ/bit differences but the increased latency which prevents the CPU from entering/exiting states as often. Nothing is instant, and you are increasing that part, meaning more time in high power active and less time in low power idle, and for modern processors it works even in moderate usage scenarios.

I have seen this failure over and over and over again. This is why earlier versions of Turbo and per core frequency changes have failed - latency.

Skymont can be very power efficient, but it has to be in a same format as Lunarlake did. The regular E cores are just like predecessors and meant for extra performance, and zero about battery life, meaning it's up to the LPE cores.

Dell on the leaked roadmap shows that they themselves also agree it's NOT the same as Lunarlake and will be significant regressions in that regard. By focusing on finances again Intel is potentially losing much more by not stemming the increasing ARM on PC threat.

 

[ajsdkflsdjfio]

I said on package memory(not enough alone but LNL has PHY optimizations) alone is enough for 0.5-1W reduction which is substantial.

20-30% reduction makes it closer to Meteorlake than Lunarlake, and that's a huge amount. You are going from 15 hours to 10-12 hours.

What I mean by "20-30% especially in specific light-work loads", is that a 20-30% difference is mostly going to be seen only in very specific scenarios in which the workload is extremely light and constant for the 12hr+ test. In actual use where you have a mix of workloads, PTL will likely continue to keep up with LNL much better than MTL could. Either way, battery life is a separate discussion from power-efficiency. Zen 4/5 has been much more power efficient than ADL/RPL but if you look at idle power consumption and certain light workloads like media consumption tests, ADL/RPL is superior meaning battery life on those CPUs would be better.

Regardless, if your definition of power efficiency is only dictated by 12HR straight of light web-browsing or streaming 1080p video then our disagreement is just about semantics. When I talk about efficiency i'm not just talking about battery life in ideal scenarios I'm talking about power/perf curve behaviors, of which PTL will be incredibly similar to lunar-lake since they possess similar core architectures and core layouts and 1-2W of extra power from lack of MoP, PMICs, and possibly inter-core power loss is going to be negligible in a lot of circumstances. Limit Panther/Lunar/Meteor to 17-30W and run any workload and you'll see that PTL and LNL are off in their own world with MTL far behind.

I told you the inter-die communication effects aren't necessarily about pJ/bit differences but the increased latency which prevents the CPU from entering/exiting states as often. Nothing is instant, and you are increasing that part, meaning more time in high power active and less time in low power idle, and for modern processors it works even in moderate usage scenarios.

If you are talking about increased latency in regards to having LPE off the compute tile, like I mentioned it's not a given that the LPE cores are actually off the compute tile same as it was on lunar-lake, and even if they were, they might be able to do what Intel originally intended them to do on meteor-lake, which is to be power enough to actually take over during light-workloads and completely turn off the compute tile. In that case, panther-lake might actually have a battery advantage over Lunar-lake regardless of increased latency, but I doubt that's what they are trying to do this time around.

Skymont can be very power efficient, but it has to be in a same format as Lunarlake did. The regular E cores are just like predecessors and meant for extra performance, and zero about battery life, meaning it's up to the LPE cores.

PTL-U has 4 LPE cores same as Lunar lake does, and it's possibly on the compute tile too, so IDK what you mean.

By focusing on finances again Intel is potentially losing much more by not stemming the increasing ARM on PC threat.

I hope you aren't thinking of Snapdragon or M-series chips as a threat to the x86 PC market. The only possible contender might be Nvidia, but that's up in the air since we literally know nothing about their plans.

Dell on the leaked roadmap shows that they themselves also agree it's NOT the same as Lunarlake and will be significant regressions in that regard.

If you are referring to the leaked roadmap in May, I really don't see how you can extrapolate that PTL-U will not be competitive in power efficiency compared to LNL. In the first place it doesn't seem to be entirely accurate since it shows LNL XPS 13 being introduced in 2026 Q1 at the same time as the XPS-14 with PTL which makes no sense since the XPS13 already has lunar lake in it. Also they have meteor-lake being used in lieu of ARL-H and PTL-H which makes zero sense if you watched a single review of MTL. So yea, I really wouldn't put any weight into anything on that roadmap.

Besides, Lunar-lake likely is still superior to PTL-U when it comes to battery life and graphics, while offering competitive perf/power to PTL-U. So I would not be surprised if they kept using lunar-lake in their premium thin & lights post panther lake. Although personally as a consumer, I would much rather purchase a panther lake laptop with slightly less battery life

 

[511]

PTL-U will be similar to LNL in that the SoC design will be overhauled to more closely resemble LNL, dropping the MTL/ARL design. But this is also true of PTL-H.

PLT-U differs from LNL in that it's not targeting the premium thin-and-light segment. It's focusing on the segment that U has always focused on: Lower cost.

PTL-U will be lower cost with no MoP (which was not implemented for performance but for idle power consumption). It'll be lower cost with less use of PMICs. It'll be lower cost with a smaller iGPU relative to LNL. It's NPU won't be any noticeably different than PTL-H. It's going to have worse idle efficiency than LNL and will probably have a gaming regression as well. It's not a 1:1 successor. It's more of a successor to "ARL-U", and any similarities it shares with LNL is true for the entire PTL family.

If it's GPU is 2X of the iGPU in ARL it is closer to 8 Core MTL iGPU

 

[MoistOintment]

If it's GPU is 2X of the iGPU in ARL it is closer to 8 Core MTL iGPU

PTL-U won't offer the alleged 12Xe3 iGPU that PTL-H will have. PTL-H will see an improvement in iGPU performance vs LNL, but PTL-U will see a regression.

What makes LNL so unique and well loved is its light load efficiency. That will see a regression in PTL-U.

12Xe3 iGPU is allegedly 100% faster than as ARL-H iGPU. LNL iGPU is 60% faster than ARL-H iGPU. PTL-U has a 4 Xe3 iGPU.

 

[dttprofessor]

Intel can offer every combination within CPU tile& GPU tile, if it have mass market.

 

[mikk]

12Xe3 iGPU is allegedly 100% faster than as ARL-H iGPU. LNL iGPU is 60% faster than ARL-H iGPU. PTL-U has a 4 Xe3 iGPU.

60% at what power? Peak performance shouldn't differ much between both. And what is your source by the way?

 

[MoistOintment]

Intel can offer every combination within CPU tile& GPU tile, if it have mass market.

Sure and that's true for a lot of things. Intel could offer an 8+32 ARL-S chip - but that would require another die to be made for a low volume desktop part that'll never recoup that cost. Intel could make a desktop ARL die that's 8+0 and replaces the E core die space with more L3. Intel could add a BMG tile to ARL-H. Intel could do a lot of things. None of these are technical restrictions, but business / product segmentation / binning decisions. At the end of the day, these products need to be profitable, and there's a reason why LNL is a separate product segment from the U series. They target different market segments.

LNL is more expensive to make than ARL-U. ARL-U will be a low cost chip that'll make up the bulk of the CU200 series volume and will be put in tons of corporate issues laptops.

Same will be true of PTL-U. It'll the low cost PTL product and will reduce costs vs LNL at the expense of light load and idle efficiency.

 

[MoistOintment]

60% at what power? Peak performance shouldn't differ much between both. And what is your source by the way?

Source for what? ARL-H is using functionally the same iGPU tile as MTL-H, and we already have the benchmarks out to compare.

 

[dttprofessor]

Source for what? ARL-H is using functionally the same iGPU tile as MTL-H, and we already have the benchmarks out to compare.

ACM+XMX

 

[Philste]

12Xe3 iGPU is allegedly 100% faster than as ARL-H iGPU. LNL iGPU is 60% faster than ARL-H iGPU.

Wut? LNL is like 30-35% faster than MTL in Games and less in 3dMark. Rumor says that ARL-H is ~20% faster than MTL (It's Xe+ vs Xe). So actually decently close to LNL. PTL-H is said to be ×2 in 3dMark So that means ~8k Timespy for me.

 

[MoistOintment]

Wut? LNL is like 30-35% faster than MTL in Games and less in 3dMark. Rumor says that ARL-H is ~20% faster than MTL (It's Xe+ vs Xe). So actually decently close to LNL. PTL-H is said to be ×2 in 3dMark So that means ~8k Timespy for me.

And if PTL-U has 1/3 the Xe core count, where does that leave PTL-U vs LNL?

 

[511]

PTL-U won't offer the alleged 12Xe3 iGPU that PTL-H will have. PTL-H will see an improvement in iGPU performance vs LNL, but PTL-U will see a regression.

What makes LNL so unique and well loved is its light load efficiency. That will see a regression in PTL-U.

12Xe3 iGPU is allegedly 100% faster than as ARL-H iGPU. LNL iGPU is 60% faster than ARL-H iGPU. PTL-U has a 4 Xe3 iGPU.

I know all that and that's why I said it will match MTL IGPU and ARL iGPU is 20-25% faster than MTL so it still is in the same ballpark but 12Xe3 iGPU is 2X faster than LNL which is already at Strix Point Level (Hamperd by Drivers in unsupported games)

 

[511]

And if PTL-U has 1/3 the Xe core count, where does that leave PTL-U vs LNL?

It will definitely have higher clocks like 3Ghz+ they will just make it more clock speed but anywhere around 70-75% of LNL iGPU

 

[MoistOintment]

It will definitely have higher clocks like 3Ghz+ they will just make it more clock speed but anywhere around 70-75% of LNL iGPU

Right. PTL-H will have a good iGPU. Not disputing that.

Just pointing out that PTL-U will have idle / light load efficiency and iGPU performance regressions vs LNL, so it's not a 1-1 replacement.
Which is unfortunate because PTL-H will not be as ideal for very low power gaming, such as in handhelds as it'll have too many CPU cores and too high of a TDP for that form factor.

 

[SiliconFly]

.

 

[511]

Right. PTL-H will have a good iGPU. Not disputing that.

Just pointing out that PTL-U will have idle / light load efficiency and iGPU performance regressions vs LNL, so it's not a 1-1 replacement.
Which is unfortunate because PTL-H will not be as ideal for very low power gaming, such as in handhelds as it'll have too many CPU cores and too high of a TDP for that form factor.

Yeah but if it performs 70-75% of LNL device I guess it's fine for $800 laptops don't you think ?

I am also hearing rumors of ARL-H being replaced very quickly this year with panther lake which I guess makes sense they will be in full force with it to ramp 18A and get rid of ARL-H as quickly as possible it takes around 15 Weeks for N2 Wafer to get dabbed so I assume 18A will be even more considering it will have more layers than N2

 

[SiliconFly]

.

 

[511]

It will be replaced in under 9 months though which is relatively short lived imo

 

[Philste]

And if PTL-U has 1/3 the Xe core count, where does that leave PTL-U vs LNL?

Ofc PTL-U will have a slower iGPU. But it won't be in Laptops that cost a minimum of 1400€

 

[MarkizSchnitzel]

So that means LNL does NOT have a successor after all. Different price bracket.
But also, LNL laptops can already be found for <1000€.

 

[SiliconFly]

.

 

[mikk]

Source for what? ARL-H is using functionally the same iGPU tile as MTL-H, and we already have the benchmarks out to compare.

Yes source for your 60% claim and power level. Same functionality doesn't tell much. Mobile graphics is power limited. Performance can vary a lot on a graphics with same functionality depending on different things like process node, CPU efficiency, bandwidth, power budget.

 

[SiliconFly]

.

 

[MoistOintment]

Yeah but if it performs 70-75% of LNL device I guess it's fine for $800 laptops don't you think ?

I am also hearing rumors of ARL-H being replaced very quickly this year with panther lake which I guess makes sense they will be in full force with it to ramp 18A and get rid of ARL-H as quickly as possible it takes around 15 Weeks for N2 Wafer to get dabbed so I assume 18A will be even more considering it will have more layers than N2

Of course. We'll order a few hundred PTL-U Dell Pro's or whatever they're calling the Latitude 5000 series replacement. They'll be an improvement over the Core 100U's we're currently deploying, which are already more than sufficient for office work.

It will be replaced in under 9 months though which is relatively short lived imo

As for the quick replacement: I'm also doubtful. Maybe one or two PTL laptops in December to claim 2025 launch date metrics were met - but expect most PTL-H laptops to be announced / launched at next year's CES.

At this point, I'm starting to feel that upcoming PTL-H is neither a true successor to ARL-H nor a true successor to LNL. Appears to be somewhere in between.

PTL-H is a replacement to ARL-H. PTL-U is a replacement to ARL-U. Packaging changes (improvements) don't change the market segments they're operating in.

 

[SiliconFly]

.