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.

 

[coercitiv]

"I'm playing both sides so that I always come out on top"

Both sides?! This ain't no scrub!

https://twitter.com/NVIDIAGPU_

 

[TESKATLIPOKA]

Both sides?! This ain't no scrub!

https://twitter.com/NVIDIAGPU_

This is one hardcore guy for sure.

 

[SiliconFly]

Maybe It's the same guy with 2 accounts.

Ya. Looks like the same guy.. just having his share of fun!

 

[SiliconFly]

In no universe is ARL going to be 60% faster than MTL.

Also I don't think MTL is expected to hit 5GHz when even MTL-H45 ES2 caps at 4.5GHz 1T (but I will be clear - I don't know MTL QS targets or anything, I'm guessing here).

https://twitter.com/x/status/1664375853021802496

I think Arrow Lake is going to be a great generation, but 60% is just too far.

Actually it may not be totally impossible.

60% ST IPC increase is actually impossible. We all know that. Should be somewhere between 10% to 20%.

But MT is a different beast altogether. ARL may get 8+32 cores compared to MTL 6+16 cores. Thats 2 extra P-cores & 16 extra E-cores for a total of 18 extra cores compared to MTL. 60% MT performance increase maybe possible for ARL. I'm not saying it's gonna happen. All I'm saying is there's a possibility due to the 18 extra cores.

(Correction: MTL 6+8)

 

[DrMrLordX]

But MT is a different beast altogether. ARL may get 8+32 cores compared to MTL 6+16 cores.

6c+16e Meteor Lake may as well not exist.

 

[A///]

"I'm playing both sides so that I always come out on top"

it's pride month, give the lad a break.

 

[coercitiv]

ARL may get 8+32 cores compared to MTL 6+16 cores.

The current participants in this discussion don't even know what they're talking about anymore. It's both scary and amusing to see you debating the possibility of ARL beating MTL by 60% when the original discussion started with the observation that MTL appears to have 60% higher base clocks than equivalent Raptor Lake parts. The debate was about Intel 7 vs Intel 4 and probably the advantages of DLVR among other reasons. Nothing to do with Arrow Lake.

So let's stop this ARL 8+32 60% MT performance uplift daydream before WCCFTECH quotes you as a source in their new "leak" and then some of us have to debunk that article as it gets quoted again and again all over the forum.

 

[DrMrLordX]

The current participants in this discussion don't even know what they're talking about anymore. It's both scary and amusing to see you debating the possibility of ARL beating MTL by 60% when the original discussion started with the observation that MTL appears to have 60% higher base clocks than equivalent Raptor Lake parts. The debate was about Intel 7 vs Intel 4 and probably the advantages of DLVR among other reasons. Nothing to do with Arrow Lake.

So let's stop this ARL 8+32 60% MT performance uplift daydream before WCCFTECH quotes you as a source in their new "leak" and then some of us have to debunk that article as it gets quoted again and again all over the forum.

So wait does that mean Arrow Lake will have 156% higher performance than Raptor Lake?

 

[AMDK11]

I'm curious if the leak about LionCove that it has an 8-way x86 decoder and ROB 700+ will work.

I'm also curious if LunarLake, like ArrowLake, is also based on LionCove cores.

 

[Henry swagger]

I'm curious if the leak about LionCove that it has an 8-way x86 decoder and ROB 700+ will work.

I'm also curious if LunarLake, like ArrowLake, is also based on LionCove cores.

Which leak ? .. give a link

 

[Geddagod]

Which leak ? .. give a link

It's a Raichu leak. Too lazy to dig it out on twitter (just do advanced search on twitter and type in the keywords if you want to find it yourself) but it's what he is saying .
It's a little absurd, considering the 6 wide decode of GLC is already clock gated so hard, and the size of all the other structures in the core would have to have ballooned out as well... it just seems like a massive jump.
But afaik Raichu and one other AMD leaker whose name I don't recall are the only leakers to present any 'core specific' type leaks (stuff other than just IPC/clockspeeds) so there aren't exactly any other leaks about LNC specifications/size.
And ye LNL apparently uses the same core architecture, lion cove, as ARL. Which makes sense, since both ARL and LNL are slated for a 2024 launch according to Intel themselves.

 

[A///]

this. i had this bookmarked because you spent too much time chatting about it.

https://twitter.com/x/status/1588767528360501248

 

[Henry swagger]

this. i had this bookmarked because you spent too much time chatting about it.

https://twitter.com/x/status/1588767528360501248

Thanks... uhm lion cove will have the same decode length and robs as the apple m1 🤔💻

 

[A///]

Thanks... uhm lion cove will have the same decode length and robs as the apple m1 🤔💻

don't thank me, thank gedda for remembering it. every one needs a cheer leader, even intel. zen 5 may get 8 decode too

 

[Geddagod]

don't thank me, thank gedda for remembering it. every one needs a cheer leader, even intel.

Didn't I literally just ask you to stop name calling like a couple days ago? Idk why you are trying to stick me with the 'Intel cheer leader' shtick lmao. Idk if you were personally insulted I think ARL could be using TSMC 3nm or something, since that's when you started insulting me, but cmon.
Especially since I haven't even been the most optimistic Intel person in this sub. The only 'outs' you can call me out on is believing Intel 4 is not going to differ on paper vs in products, and GNR using LNC rather than RWC+.

this. i had this bookmarked because you spent too much time chatting about it.

Like I wasn't even going to comment on this at first, but like dude, I talked about LNC going 8 wide with a massive 700 ROB like 3 times, and I'm pretty sure the only time I brought it up unprompted was to ask Exist50 on the viability of those specifications in question.
I mean I don't want to block you, but I don't even know what I did to prompt this annoying name calling. It's not even like I'm the most 'Intel biased' here on this forum, since there's Witkekin, and a guy whose username literally is IntelUser200 or something, so I'm guessing it has to be personal.

 

[A///]

Didn't I literally just ask you to stop name calling like a couple days ago? Idk why you are trying to stick me with the 'Intel cheer leader' shtick lmao. Idk if you were personally insulted I think ARL could be using TSMC 3nm or something, since that's when you started insulting me, but cmon.
Especially since I haven't even been the most optimistic Intel person in this sub. The only 'outs' you can call me out on is believing Intel 4 is not going to differ on paper vs in products, and GNR using LNC rather than RWC+.

Like I wasn't even going to comment on this at first, but like dude, I talked about LNC going 8 wide with a massive 700 ROB like 3 times, and I'm pretty sure the only time I brought it up unprompted was to ask Exist50 on the viability of those specifications in question.
I mean I don't want to block you, but I don't even know what I did to prompt this annoying name calling. It's not even like I'm the most 'Intel biased' here on this forum, since there's Witkekin, and a guy whose username literally is IntelUser200 or something, so I'm guessing it has to be personal.

it's not name calling if it's a compliment. intel's in a sticky situation at the moe and you are the only other person on this forum aside from me who has some amount of faith in them to turning around their ship. if you thought I was being a dick that was not my intention. there are other people who seldom post who are deserving of such behaviour from me.

I mean I don't want to block you, but I don't even know what I did to prompt this annoying name calling. It's not even like I'm the most 'Intel biased' here on this forum, since there's Witkekin, and a guy whose username literally is IntelUser200 or something, so I'm guessing it has to be personal.

ahem you forgot one of santa's helpers bless his heart. never seen someone do the equivalent of sticking their head in a ice box so many times.

 

[Geddagod]

it's not name calling if it's a compliment. intel's in a sticky situation at the moe and you are the only other person on this forum aside from me who has some amount of faith in them to turning around their ship. if you thought I was being a dick that was not my intention. there are other people who seldom post who are deserving of such behaviour from me.


ahem you forgot one of santa's helpers bless his heart. never seen someone do the equivalent of sticking their head in a ice box so many times.

I apologize then, I always saw it as an insult and was getting repeatedly miffed

 

[A///]

I apologize then, I always saw it as an insult and was getting repeatedly miffed

no worries. hopefully I don't eat crow 4-5 years from now if intel has not yet recovered. I have faith they will but many don't. can't blame them. people felt the same when amd was hammering them 20 years ago even with the way intel was showering odms with bribe money. core was their big turn around moment and they were hot as fire until ryzen dropped. first gen wasn't mind blowing but it was a % of intel's performance plus many more cores for cheaper. the 3950x when it launched was one the first wooden stakes to intel's heart. amd's moves along with the next gen caused intel's hedt platform to become almost useless overnight despite not having remotely the same feature or io parity. 3rd gen tr was impressive but my jaw fell the wayside when they debuted the 64 core model years ago and the pricing. id never have bought it but that kind of power for so little compared to the competition blew minds. tr7000 should be impressive. intel better hide.

 

[SpudLobby]

So is it looking like Meteor Lake doesn't actually have any efficiency improvements in the most direct sense of active workloads? The PL1 and PL2 frequencies in these engineering samples don't sound great and I'm not expecting major IPC gains. Is it actually possible the 20% more perf and 40% lower power (iso-frequency, iso-arch) totally trail off in the upper clock ranges due to the new process node? Would be seriously disappointing if MTL vs a similar ADL config at 2-3.5GHz brings negligible power and energy savings.


Idle draw if improved via the LP E cores should be big of course since you can always set a different power plan for the main cores, then you'd basically presumably have something a bit closer to Tiger Lake in terms of runtime. But put that aside.

 

[coercitiv]

So is it looking like Meteor Lake doesn't actually have any efficiency improvements in the most direct sense of active workloads? The PL1 and PL2 frequencies in these engineering samples don't sound great and I'm not expecting major IPC gains. Is it actually possible the 20% more perf and 40% lower power (iso-frequency, iso-arch) totally trail off in the upper clock ranges due to the new process node? Would be seriously disappointing if MTL vs a similar ADL config at 2-3.5GHz brings negligible power and energy savings.

Having a process that is both 40% lower power at 3Ghz and say... 10% higher power at 5Ghz+ is possible. For a mobile device such a v/f curve is still great as long as the CPU maker is not under pressure to offer ST clocks beyond the efficiency threshold (relative to the older node and older gen products).

That being said, if MTL base clocks increase @ ISO TDP is true and subject to the same rules as older gens, then the big jump in base clocks (~60%) indicates strong efficiency improvements in active workloads. At this point in time we have little reason to believe MTL won't deliver overall efficiency improvements, even when maintaining a cautious stance.

 

[mikk]

So is it looking like Meteor Lake doesn't actually have any efficiency improvements in the most direct sense of active workloads? The PL1 and PL2 frequencies in these engineering samples don't sound great and I'm not expecting major IPC gains. Is it actually possible the 20% more perf and 40% lower power (iso-frequency, iso-arch) totally trail off in the upper clock ranges due to the new process node? Would be seriously disappointing if MTL vs a similar ADL config at 2-3.5GHz brings negligible power and energy savings.


Idle draw if improved via the LP E cores should be big of course since you can always set a different power plan for the main cores, then you'd basically presumably have something a bit closer to Tiger Lake in terms of runtime. But put that aside.

PL1 not improved? PL1 goes up from like 2 Ghz to 3+ Ghz. We don't really have PL2 numbers (power consumption+frequency). Comparing ES2 to a shipping CPU is invalid. MTL is not only about Intel 4, it's also about the power management. A separate voltage rail for E cores could give a nice advantage in PL2 scenarios as well.

 

[eek2121]

So is it looking like Meteor Lake doesn't actually have any efficiency improvements in the most direct sense of active workloads? The PL1 and PL2 frequencies in these engineering samples don't sound great and I'm not expecting major IPC gains. Is it actually possible the 20% more perf and 40% lower power (iso-frequency, iso-arch) totally trail off in the upper clock ranges due to the new process node? Would be seriously disappointing if MTL vs a similar ADL config at 2-3.5GHz brings negligible power and energy savings.


Idle draw if improved via the LP E cores should be big of course since you can always set a different power plan for the main cores, then you'd basically presumably have something a bit closer to Tiger Lake in terms of runtime. But put that aside.

Base frequencies of PL1/2 jumped from ~2ghz to 3.1+ ghz and you are saying that isn’t an improvement??

Note that you can’t compare h/s chips, you need to look at “P” chips.

EDIT: Just wanted to add that these chips won’t likely perform much better than non-ultra parts. If I am understanding Intel’s intent correctly, these chips are designed to offer similar performance at a lower power envelope. The primary target will be premium ultrabook type devices, devices where cooling is limited, possibly even fanless devices.

 

[SpudLobby]

Base frequencies of PL1/2 jumped from ~2ghz to 3.1+ ghz and you are saying that isn’t an improvement??

Note that you can’t compare h/s chips, you need to look at “P” chips.

EDIT: Just wanted to add that these chips won’t likely perform much better than non-ultra parts. If I am understanding Intel’s intent correctly, these chips are designed to offer similar performance at a lower power envelope. The primary target will be premium ultrabook type devices, devices where cooling is limited, possibly even fanless devices.

Right, that's my understanding as well: this is about power efficiency. I guess I was under the impression PL1 didn't move either. Uzzi seems fairly negative among others. Frankly I don't care if max ST doesn't change, but a 40% reduction in power iso-frequency in the 2-3GHz range would be huge if they can pull it off with the new node and extra cache, of course they have other management techniques on the menu namely the low power cores and shutting off the compute tile.

 

[dullard]

Is it actually possible the 20% more perf and 40% lower power (iso-frequency, iso-arch) totally trail off in the upper clock ranges due to the new process node? Would be seriously disappointing if MTL vs a similar ADL config at 2-3.5GHz brings negligible power and energy savings.

Coercitiv, mikk, and eek2121 covered it pretty well. But I do want to highlight the word 'and' in your quote. That word should be 'or'. 20% more performance or 40% lower power. We aren't getting both. Not only that, but those numbers are only accurate when the processor is run at ~0.65 V. Look at the circles in this image. Intel 4 gets either 21% more frequency at the same power OR 40% less power at the same frequency.

AnandTech Forums: Technology, Hardware, Software, and Deals

Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

www.anandtech.com

But, now look at the square dots. At 0.85V and 3 GHz frequency, Intel 4 was only ~35% less power than Intel 7.

As you keep increasing the voltage, the possible frequencies get higher but the curves start to bend sharply upwards to even higher power. Extrapolate this out to 5 GHz or 6 GHz and the power could be massive. The Intel 4 node isn't optimized for high frequency use. The data isn't shown in the graph but it could be possible that MTL is just not going to perform well at ADL frequencies. Wait a few more months for Arrow Lake if that is your need.

 

[Khato]

EDIT: Just wanted to add that these chips won’t likely perform much better than non-ultra parts. If I am understanding Intel’s intent correctly, these chips are designed to offer similar performance at a lower power envelope. The primary target will be premium ultrabook type devices, devices where cooling is limited, possibly even fanless devices.

I'd disagree that MTL isn't going to perform much better than the non-ultra parts... just not on the CPU side The major improvements for MTL are efficiency as you note and graphics performance. Which makes sense as those are the two areas where they're lagging competition.