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

[Tigerick]

Wildcat Lake (WCL) Specs

Intel Wildcat Lake (WCL) is upcoming mobile SoC replacing Raptor Lake-U. 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 Q1 2026.

		Intel Raptor Lake U	Intel Wildcat Lake 15W?	Intel Lunar Lake		Intel Panther Lake 4+0+4
Launch Date		Q1-2024	Q2-2026	Q3-2024		Q1-2026
Model		Intel 150U	Intel Core 7	Core Ultra 7 268V		Core Ultra 7 365
Dies		2	2	2		3
Node		Intel 7 + ?	Intel 18-A + TSMC N6	TSMC N3B + N6		Intel 18-A + Intel 3 + TSMC N6
CPU		2 P-core + 8 E-cores	2 P-core + 4 LP E-cores	4 P-core + 4 LP E-cores		4 P-core + 4 LP E-cores
Threads		12	6	8		8
Max Clock		5.4 GHz	?	5 GHz		4.8 GHz
L3 Cache		12 MB		12 MB		12 MB
TDP		15 - 55 W	15 W ?	17 - 37 W		25 - 55 W
Memory		128-bit LPDDR5-5200	64-bit LPDDR5	128-bit LPDDR5x-8533		128-bit LPDDR5x-7467
Size		96 GB		32 GB		128 GB
Bandwidth				136 GB/s
GPU		Intel Graphics	Intel Graphics	Arc 140V		Intel Graphics
RT		No	No	YES		YES
EU / Xe		96 EU	2 Xe	8 Xe		4 Xe
Max Clock		1.3 GHz	?	2 GHz		2.5 GHz
NPU		GNA 3.0	18 TOPS	48 TOPS		49 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.

 

[511]

speculate one reason that they keep having these cycles is because they change their CEOs too much. They'd be much better off with a good CEO that stays there for decades, like Jensen.

The Intel Trinity thought differently they allowed everyone to have a chance at Position Rob Noyce/Gordon Moore retired in relatively short time as CEO compared to other folks Andy Grove had health issues.

 

[511]

8% over 18A puts it a hair faster than N3p.

18A is in the same ballpark at perf/watt with N3 at lower ranges it's at higher frequency it starts to flat out they need to fix it and improve the fmax and it will be close to N2.

 

[DavidC1]

18AP is ahead of N3P ish closer to N2

Individual transistor performance and clockspeed has broken it's relation a long time ago anyway.

Icelake's 10nm was slower than 10nm SF and Intel 7, but the smaller transistor offered lower capacitance meaning overall lower power. Also according to the 18A presentation it's better performing at 1.1V, then 0.75V. Meaning it does better for desktop and laptop processors, the opposite of what you are saying. That's also confirmed by Intel's presentation that says they won't have "mobile" variants until 18A-P.

The Intel Trinity thought differently they allowed everyone to have a chance at Position Rob Noyce/Gordon Moore retired in relatively short time as CEO compared to other folks Andy Grove had health issues.

Yea and look at Nvidia... and look at Intel.

8% over 18A puts it a hair faster than N3p.

Intel 7 is roughly equal to N7 in drive current. If you extrapolate that out here's what we get:

Intel 7/N7: 1x
Intel 4: 1.2x
Intel 3: 1.18x
18A: 1.15x

1.628x

N5: 1.15
N3E: 1.18
N3P: 1.05
N3X: 1.05
N2P(vs N3E): 1.18x

1.496x for N3X, 1.6x for N2P

Intel 18A Process Node Offers 25% Higher Frequency At ISO & 36% Lower Power At Same Frequency Versus Intel 3, Over 30% Density

Intel has detailed its next-gen 18A process node, which will be the replacement for its Intel 3 node, offering improved clock & voltage scaling. Intel 18A Process Node Offers Improved Performance, Efficiency & Density Versus Intel 3 Node At the 2025 Symposium on VLSI Technology and Circuits...

semiwiki.com

18A also achieves 25% gain at 1.1V, which is basically Turbo clocks. It's 18% at 0.75V. The graph says it achieves over 15% over Intel 3.

What's achievable on a transistor level doesn't necessarily scale to clockspeeds at the whole chip level. The better transistors allow you to do other things such as different circuit designs. Intel's 22nm process had great gains over 32nm, especially for Atom with 37% gains. However, with Ivy Bridge, overclock capability degraded noticeably. It took them two years later with Haswell Refresh and 4790K to get better clocks than 32nm.

 

[511]

Yea and look at Nvidia... and look at Intel.

Intel is fair bit older than Nvidia and yeah they should have chosen a good CEO after Craig Barett last capable CEO before PG but PG didn't actually solved the core problem the rot that infested during PSO-Swan Era

ntel 7 is roughly equal to N7 in drive current. If you extrapolate that out here's what we get:

Intel 7/N7: 1x
Intel 4: 1.2x
Intel 3: 1.18x
18A: 1.15x

1.628x

N5: 1.15
N3E: 1.18
N3P: 1.05
N3X: 1.05
N2P(vs N3E): 1.18x

1.496x for N3X, 1.6x for N2P

It's not showing up in designs though 18AP is behind N2P

 

[DavidC1]

Intel is fair bit older than Nvidia and yeah they should have chosen a good CEO after Craig Barett last capable CEO before PG but PG didn't actually solved the core problem.

Craig made manufacturing strong(his background) but he didn't do good on products. That's why I call him the "sane" Kraznich.

It's not showing up in designs though 18AP is behind N2P

I added the part about 22nm vs 32nm:

Intel's 22nm process had great gains over 32nm, especially for Atom with 37% gains. However, with Ivy Bridge, overclock capability degraded noticeably. It took them two years later with Haswell Refresh and 4790K to get better clocks than 32nm.

It's not that simple as having faster transistors for high clocks anymore. Actually this thinking is a decade late. It stopped after 45nm generation.

What's degraded is Intel's engineering too. When layoffs happen, the hidden parts that really kept everything stable falls apart. That's why they are behind others in memory adoption, or cache latencies and sizes. The "boring" stuff is what matters for a product because the 5-10% is the difference between equal/lose or being at the top.* Arrowlake sucking in games despite having good foundation(Cores), is because it lacks the engineering necessary to smooth out the rough edges. We'll see whether the recent round of layoffs by both Tan and Pat will have such negative impacts in the future.

*I was fascinated with Itanium, and Itanium 2 had fantastic engineering. First chip with a massive on-die cache, that was very fast, and also very compact. The 0.13u successor was amazing in that regard as well. If you go older, their chipsets were usually the top performing. When they went integrated memory controller, they were way ahead of AMD. And IDF always talked about new memory, PCIe, and USB standards, among others. They pushed the industry forward.

 

[Meteor Late]

Intel will use N2P for the higher end desktop parts. That tells me all I need to know in regards to 18A-P (and 18A by extension) clock speed potential, or lack of.

 

[DavidC1]

Intel will use N2P for the higher end desktop parts. That tells me all I need to know in regards to 18A-P (and 18A by extension) clock speed potential, or lack of.

Again, 22nm vs 32nm.

 

[dullard]

Not sure how I'm misinterpreting this, "begin production ramp of Intel 18A in these products in the first half of '25 for product release in the middle of next year." How should I read that as anything other than exactly what it says?

If you are confused by a long sentence, then you can remove the words that just add flair. For example, "the cat with black and white fur has catnip" can be reduced to "the cat with black and white fur has catnip" or "the cat has catnip". That message is still true. The cat has catnip regardless of the cat's color. But, it is nice extra information to know the cat's color.

Now lets try that with your message.

“Intel's CEO, Pat Gelsinger, has confirmed that the upcoming 18A process of the Panther Lake CPU generation is on schedule for a mid-2025 release, which aligns with the initial projection.”

reduces to

“Intel's CEO, Pat Gelsinger, has confirmed that the upcoming 18A process of the Panther Lake CPU generation is on schedule for a mid-2025 release, which aligns with the initial projection.”

or

“Intel's CEO has confirmed that the upcoming 18A process is on schedule for a mid-2025 release.”

You know, the whole 18A process that Intel tried to sell as a foundry to other companies. Doesn't look like Intel succeeded in that quest. But it was tested by several companies in mid-2025. The new CEO is focusing on 14A for external companies.

 

[511]

Intel will use N2P for the higher end desktop parts. That tells me all I need to know in regards to 18A-P (and 18A by extension) clock speed potential, or lack of.

Fmax shouldn't be the comparison Cause Intel 7 has crazy Fmax

 

[adroc_thurston]

Individual transistor performance and clockspeed has broken it's relation a long time ago anyway.

Idk it works for TSM.
You get a faster node and (you're not gonna believe that!) go faster.

 

[DavidC1]

Idk it works for TSM.
You get a faster node and (you're not gonna believe that!) go faster.

So by that logic N3B is 5% slower than Intel 7.

 

[Win2012R2]

I agree that Larrabbee was a failure. But, he should have been reprimanded instead

Why? He obviously had to use x86 for it and nobody in this Universe could have made it work better than custom stuff AMD/Nvidia were doing without any worry for backwards compatibility other than defined software APIs.

Also Many Cores concept from that project came up pretty handy when core counts gone up a lot.

 

[adroc_thurston]

So by that logic N3B is 5% slower than Intel 7.

It's however much faster than N4 if that's what you're going for.

 

[DavidC1]

Why? He obviously had to use x86 for it and nobody in this Universe could have made it work better than custom stuff AMD/Nvidia were doing without any worry for backwards compatibility other than defined software APIs.

Also Many Cores concept from that project came up pretty handy when core counts gone up a lot.

Because Larrabbee did fail. There does need to be some form of punishment. What I am saying is firing was the wrong move.

 

[Hitman928]

If you are confused by a long sentence, then you can remove the words that just add flair. For example, "the cat with black and white fur has catnip" can be reduced to "the cat with black and white fur has catnip" or "the cat has catnip". That message is still true. The cat has catnip regardless of the cat's color. But, it is nice extra information to know the cat's color.

Now lets try that with your message.

“Intel's CEO, Pat Gelsinger, has confirmed that the upcoming 18A process of the Panther Lake CPU generation is on schedule for a mid-2025 release, which aligns with the initial projection.”

reduces to

“Intel's CEO, Pat Gelsinger, has confirmed that the upcoming 18A process of the Panther Lake CPU generation is on schedule for a mid-2025 release, which aligns with the initial projection.”

or

“Intel's CEO has confirmed that the upcoming 18A process is on schedule for a mid-2025 release.”

You know, the whole 18A process that Intel tried to sell as a foundry to other companies. Doesn't look like Intel succeeded in that quest. But it was tested by several companies in mid-2025. The new CEO is focusing on 14A for external companies.

Now do that with his actual quote, which I already posted twice, not the butchered one.

"Furthermore, our lead products, Clearwater Forest and Panther Lake are already in fab, and we expect to begin production ramp of Intel 18A in these products in the first half of '25 for product release in the middle of next year.”

 

[Win2012R2]

Because Larrabbee did fail. There does need to be some form of punishment. What I am saying is firing was the wrong move.

They aimed it at HPC where it did score some wins in supercomputers of the time (had problems delivering tho), fundamentally it's failed because Intel brought x86 to a fight that involves custom stuff with no need to maintain backwards compatibility (other than a few APIs), hardly Pat's fault. Pretty sure it was not funded properly too, they (Intel top brass) probably did it to prevent Pat from becoming CEO.

 

[Cheesecake16]

I am just going to post this excellent blog from Tom Forsyth about why Larrabee didn't fail nearly as bad as people think it did...
https://tomforsyth1000.github.io/blog.wiki.html#%5B%5BWhy%20didn't%20Larrabee%20fail%3F%5D%5D

 

[dullard]

Now do that with his actual quote, which I already posted twice, not the butchered one.

"Furthermore, our lead products, Clearwater Forest and Panther Lake are already in fab, and we expect to begin production ramp of Intel 18A in these products in the first half of '25 for product release in the middle of next year.”

You can't eliminate a comma and the word "and" then randomly apply the remaining part of a sentence to the first portion before the comma.

"Furthermore, our lead products, Clearwater Forest and Panther Lake are already in fab, and we expect to begin production ramp of Intel 18A in these products in the first half of '25 for product release in the middle of next year.”

Inventing your own language rules to pretend a company is behind schedule. Weird flex, man. But you do you.

 

[Hitman928]

You can't eliminate a comma and the word "and" then randomly apply the remaining part of a sentence to the first portion before the comma.

"Furthermore, our lead products, Clearwater Forest and Panther Lake are already in fab, and we expect to begin production ramp of Intel 18A in these products in the first half of '25 for product release in the middle of next year.”

Inventing your own language rules to pretend a company is behind schedule. Weird flex, man. But you do you.

You just eliminated the subject of the sentence to try to twist it to fit what you want it to say, lol. You can keep tilting at windmills, but I think it’s pretty apparent to everyone else what the sentence says.

Now I remember why I stopped posting here much, too many people entrenched in camp a or camp b that they will fight in the weirdest ways to try and make their side not look bad, even over the simplest of things.

 

[LightningZ71]

Intel will use N2P for the higher end desktop parts. That tells me all I need to know in regards to 18A-P (and 18A by extension) clock speed potential, or lack of.

It could also be an indictment of the yields (parametric or just working chips) AND capacity that they have on 18a. If they just don't have sufficient yielded capacity to supply the market with their products that use larger CCDs, no matter how "good" 18a might actually be, it may just make business sense to go to TSMC for more capacity.

 

[OneEng2]

Arrow Lake I would argue was somewhat of a comeback for Intel. Outside of gaming, Arrow Lake is a very competitive product and truth be told Arrow Lake is acceptable for gaming, it's just that AMD is demonstrably better.

I'm very much looking forward to Nova Lake and Zen 6. Hopefully they will both be successful and we can look forward to these tech giants going head-to-head for a couple more generations before some new paradigm changes the playing field.

I agree.

People like to bash it because of gaming performance, but overall ARL is a pretty good product from a consumer's perspective.

Where I feel it falls short is in the cost to produce it (same for LNL). I suspect Intel is paying a pretty penny for each processor compared to AMD's Zen 5 on N4P.

 

[511]

w

Where I feel it falls short is in the cost to produce it (same for LNL). I suspect Intel is paying a pretty penny for each processor compared to AMD's Zen 5 on N4P.

you do realize the part of ARL that is on N3B is Compute tile and the IO/SOC is on N6 which is dirt cheap the N5 iGPU is 23mm2 so the only costly stuff is the compute die and packing is somewhat expensive.

 

[dullard]

You just eliminated the subject of the sentence to try to twist it to fit what you want it to say, lol. You can keep tilting at windmills, but I think it’s pretty apparent to everyone else what the sentence says.

Now I remember why I stopped posting here much, too many people entrenched in camp a or camp b that they will fight in the weirdest ways to try and make their side not look bad, even over the simplest of things.

You missed the mark here. AMD and Intel both have great products. I'm one of the few that uses both and supports both. If neutral people make you think they are biased, then you need to look into the mirror to see the bias.

I post here when people make obviously false claims. In this case, it was you that made one.

 

[Hitman928]

You missed the mark here. AMD and Intel both have great products. I'm one of the few that uses both and supports both. If neutral people make you think they are biased, then you need to look into the mirror to see the bias.

I post here when people make obviously false claims. In this case, it was you that made one.

If you have to delete the subject of a sentence to get it to say what you want it to say, you’re not reading it truthfully.

 

[coercitiv]

People like to bash it because of gaming performance, but overall ARL is a pretty good product from a consumer's perspective.

It's a good product now, after numerous firmware updates and a considerable price drop.

When it launched, the 265K was just ~3% 2% faster than 14700K in Techpowerup's testing (performance in applications, gaming excluded). The worst offender in my book was the poor browser performance, making it slower than a 12700K in Speedometer or Jetstream. The pricing added insult to injury, as it matched the 9900X.



People bash Arrow Lake because it started on the wrong foot and Intel never did manage to shake this bad image.