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+4+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.

 

[poke01]

If you had to pick 3 application tests and/or benchmarks to get a rough indication of ST IPC (throughput) of a CPU, what would they be

Easy it’s web benchmarks. you already got two. This one is more useful for more people

 

[jdubs03]

Easy it’s web benchmarks. you already got two. This one is more useful for more people

Speedometer 3.1, Jetstream 2.2, Kraken 1.1?
Only thing about Kraken is it’s so short.

Cinebench 2026 is good too but not a web benchmark.

 

[poke01]

Speedometer 3.1, Jetstream 2.2, Kraken 1.1?
Only thing about Kraken is it’s so short.

Cinebench 2026 is good too but not a web benchmark.

Speedometer and JetStream is good enough

 

[Cardyak]

The reason I bring this up is because it would be great to create an IPC table we could use for discussion. If we could agree on the metrics and some people could even submit scores for 1 CPU, we could create such a table in short order.

I have a microarchitecture cheat sheet that lists normalised IPC performance. It also contains architectural details and diagrams going all the way back to the original Intel Pentium.

https://docs.google.com/spreadsheet...gdB9oLbTJCFwx0iFI-vUs6WFyuE/edit?usp=drivesdk

 

[DavidC1]

@Hulk

Well that's not true otherwise it would have been in Intel slide deck but it ain't they never mention IPC improvement all PTL did was fix the uncore in ARL.

It's not 2-3%, that's what we get with minor changes like Crestmont. The Cougar Cove got bigger changes. 5-10% is believable.

It's clocking lower, hence cancelling out the advantages. It needs to be 5% faster just to cancel out the loss in clocks.

 

[DavidC1]

I wonder if any of the proper reviews for PTL graphics will include Strix Halo running at comparable power. Anyone aware of existing reviews for Strix Halo graphics which aren't running at 100W+?

The comparison points in notebookcheck's brief CES PTL preview, including external monitor Cyberpunk 2077 power numbers from the full reviews - https://www.notebookcheck.net/Intel...in-our-first-gaming-benchmarks.1200743.0.html
- Radeon 8060S: 92.1 fps, 195W
- RTX 5050: 75.2 fps, 129W
- Arc B390: 44.8 fps, 60W (according to PCworld preview, full system)
- Arc 140V: 28.3 fps, 45W
- Radeon 890M: 26.5 fps, 58W

Here are better numbers:

https://www.notebookcheck.net/HP-ZBook-Ultra-G1a-14-review-Powerful-MacBook-Pro-alternative-for-work-and-game.994758.0.html

The desktop system overrates it because it uses more power hungry non-CPU components. The above is a Strix Halo laptop.

66W PL1 with 110W system power.

The 140V is 45W with 23W PL1. The 890M is 60W with 36W PL1. The TDP differences are approximately the system power difference. So according to those numbers, if Pantherlake system is at 60W, then the TDP is set to approximately 35W, similar to Strix Point. It could be set as high as 40W, because Lunarlake is lower than previous gen and AMD systems due to more efficient system, and Pantherlake inherits this.

Yes, but the Intel ARC 390 is not 60W full system but 70W CPU sustained, actualy
70W sustained PL1, 85W PL2 and 212W PL4, Computerbase also did a test, so there s a long shot from 60W full system to say 65W sustained CPU power.

Edit : That s about 2x the TDP of a LNL, so if the GPU is 2x bigger there s basicaly no improvement in perf/watt from N3 to 18A.

That's your bias talking. A system can't use only 60W when the CPU is set at 60W. 60W system is 35-40W CPU.

 

[OneEng2]

What other bet could Intel have made? 18A is their crown jewel, and Panther Lake is their best foot forward on it. The only other 'bet' would have been to go the way of AMD and Global Foundries, and marry their future products' success solely with TSMC, something the US Government would never have let happen.

At the end of the day, one has to ask themselves is Intel (now) primarily a foundry or a products company? I would argue they've been trying like heck to transition into the former, and they're hoping Panther Lake is a "good enough" effort of the latter to get them through this transition, until their Foundry is good enough to bring in the likes of Apple, NVIDIA, Qualcomm, and even AMD.

They could have (like AMD) realized they no longer have the volume and scale to keep up with a pure foundry like TSMC in the long term and went foundryless.

Instead, they bet they can keep up despite the lower volumes that TSMC enjoys..... and with the help of the US government .... they may have been correct..... so all my previous speculation about how stupid a business move this was, may end up being very smart ..... on the back of US tax payers though.

I think AMD's client business will feel more pain from increased memory prices as they are very consumer focused - consumer laptop (low enterprise presence) & DIY desktop gaming. Intel's client business being more oriented towards OEM & Enterprise is going to shield them a bit better. Win 11 refresh also favors Intel more again due to enterprise exposure.

Possibly so; however, it's hard to sell OEM stuff at razor thin margins without the high end stuff to make the profit (DC, HEDT, Gaming, etc).

This whole topic is retarded / lying reporters (and people like MLID), who have no clue "report" to other retards, and then these retards think they have some understanding.

Oh, I would say that I know plenty of pretty good sized businesses (10's of billions a year) that did rush a bunch of stuff from China and other Asian countries to get buffered up before the tariffs hit .... but still, the tariffs are VERY real and are hitting almost everyone now. I expect that unless the tariffs are overturned (might see this on Wednesday ... woot!), inflation will really take off in 2026. Companies WONT absorb the losses for very long (if at all).

NVL is a decent Tock unlike Lame Cove

I think you might be right. It might be a very decent Tock. Time will tell.

Well that's not true otherwise it would have been in Intel slide deck but it ain't they never mention IPC improvement all PTL did was fix the uncore in ARL.

... and that may be more than enough to really make the core breath.

I am sooooo happy 18A seems to be working out and Panther Lake looks like a winner for Intel!

I'm right there with you!

IPC was improved but clock was lowered. So ST is stagnant.

Make no mistake, the CPU is a flop. It’s everything else that is great.

How? This is a notebook chip. It looks pretty good to me. My only concern is the cost to produce it, not its performance in its most critical metrics for its market.

Maybe I am missing something?

 

[511]

IPC was improved but clock was lowered. So ST is stagnant.

Make no mistake, the CPU is a flop. It’s everything else that is great.

CPU is not a flop though?

 

[msj10]

IPC was improved but clock was lowered. So ST is stagnant.

Make no mistake, the CPU is a flop. It’s everything else that is great.

The biggest flop of PTL is 18A

 

[511]

I have a microarchitecture cheat sheet that lists normalised IPC performance. It also contains architectural details and diagrams going all the way back to the original Intel Pentium.

https://docs.google.com/spreadsheet...gdB9oLbTJCFwx0iFI-vUs6WFyuE/edit?usp=drivesdk

E core vs P core track record is funny in IPC

 

[511]

The biggest flop of PTL is 18A

Lmao the real data is not out yet
You need
Power
Performance
Arra
To judge a node we haven't got anything

 

[DavidC1]

Black myth wukong 1080 medium upscaling off

This actually matches Intel data against 4050S in a relative sense.

They got 51 fps for 4050 @ 60W versus PTL at 44 fps.

 

[msj10]

Lmao the real data is not out yet
You need
Power
Performance
Arra
To judge a node we haven't got anything

500 MHz clock regression on the E core. I don't need any more than that to see it's a flop.

 

[DavidC1]

I have a microarchitecture cheat sheet that lists normalised IPC performance. It also contains architectural details and diagrams going all the way back to the original Intel Pentium.

https://docs.google.com/spreadsheet...gdB9oLbTJCFwx0iFI-vUs6WFyuE/edit?usp=drivesdk

Goldmont is ~30% over Silvermont.

 

[511]

500 MHz clock regression on the E core. I don't need any more than that to see it's a flop.

It's -400 Mhz also E cores have shrinked decent bit waiting for official Spec testing

 

[DavidC1]

It's -400 Mhz also E cores have shrinked decent bit

285H is 4.5GHz, that's 500MHz.

 

[poke01]

CPU is not a flop though?

Yes it is, the ST is stagnant

 

[DavidC1]

CPU is not a flop though?

It is in a sense that Elite X2 and Zen 6 is coming. One step at a time I guess.

Pantherlake is Icelake 2025. CPU is meh, GPU is a big advancement, and the node is still immature. So 18A-P next year might be like Tigerlake and get oversized gains. The difference this time is that the iGPU was already competitive, when in Icelake it was not.

 

[Khato]

Here are better numbers:

https://www.notebookcheck.net/HP-ZBook-Ultra-G1a-14-review-Powerful-MacBook-Pro-alternative-for-work-and-game.994758.0.html

The desktop system overrates it because it uses more power hungry non-CPU components. The above is a Strix Halo laptop.

66W PL1 with 110W system power.

The 140V is 45W with 23W PL1. The 890M is 60W with 36W PL1. The TDP differences are approximately the system power difference. So according to those numbers, if Pantherlake system is at 60W, then the TDP is set to approximately 35W, similar to Strix Point. It could be set as high as 40W, because Lunarlake is lower than previous gen and AMD systems due to more efficient system, and Pantherlake inherits this.

Sure, using the same metrics from that review puts the Radeon 8060S at 80.7 fps, 110W.

Important points in my opinion are as follows:
- PTL graphics beats AMD's mainstream 128b mobile platform in both performance and efficiency.
- PTL graphics is probably roughly on par with Strix Halo when both are running at the ideal power level for PTL. Obviously Strix Halo has far more headroom due to being an absurdly expensive 256b platform. Which is also why it doesn't scale down to PTL power levels particularly well.
- PTL CPU is going to be just fine, just like LNL and ARL were just fine. Sure initial 18A didn't quite meet what the PDK promised and hence frequencies are same as LNL, but power and area are still great.

 

[jdubs03]

It is in a sense that Elite X2 and Zen 6 is coming. One step at a time I guess.

To be fair Zen 6 should be compared against NVL.

 

[DavidC1]

Sure, using the same metrics from that review puts the Radeon 8060S at 80.7 fps, 110W.

Important points in my opinion are as follows:
- PTL graphics beats AMD's mainstream 128b mobile platform in both performance and efficiency.
- PTL graphics is probably roughly on par with Strix Halo when both are running at the ideal power level for PTL. Obviously Strix Halo has far more headroom due to being an absurdly expensive 256b platform. Which is also why it doesn't scale down to PTL power levels particularly well.
- PTL CPU is going to be just fine, just like LNL and ARL were just fine. Sure initial 18A didn't quite meet what the PDK promised and hence frequencies are same as LNL, but power and area are still great.

Yea I just wanted to show why the Halo power use was so high. The desktop components use quite a bit more power, even under idle.

Strix Halo is not competition, except for the top 5% income people. The problem is less the 256-bit memory width, but the base design(motherboard, design), can't be reused from the mainstream Strix Point. AMD is also getting the unexpected benefit from the AI slop era, with people saying AMD meant it for AI first, when really, it was basically gaming and AI happened to be there. The demand would have been even lower otherwise.

I hate to say I told you so, but I told you so. I like the idea of halo iGPUs, but it falls flat on its face for practicality. People are enamored by halo iGPUs for two reasons, that iGPUs are cheap, and is low power. Halo fails:
-Because AMD/Intel has big iGPU not for charity purposes, but premium to get more money from you
-Then if the price is high, now you face against Nvidia, the one that actually has mindshare, and one that can use it's marketing and volume advantages to cancel out any theoretical cost advantages, and more
-The battery can be done well, but that depends on implementation. Strix Halo does this decently
-Load TDP is non-negotiable. More performance is more power.

It's likely Novalake-AX or whatever Intel does will fail for the same reason. The cheapest I think they can do system-wise is eDRAM like Intel did, in a same pinout and same memory standards and no extra shaders, so you are just benefitting from faster memory. You can get ~20-25% per W, and charge extra $100 on configuration settings. It's exactly what Intel did with non-Pro Iris parts with eDRAM.

 

[poke01]

Look here

Dell Inc. XPS 14 DA14260 - Geekbench

Benchmark results for a Dell Inc. XPS 14 DA14260 with an Intel Core Ultra X7 358H processor.

browser.geekbench.com

HP HP OmniBook Ultra Flip Laptop 14-fh0xxx - Geekbench

Benchmark results for a HP HP OmniBook Ultra Flip Laptop 14-fh0xxx with an Intel Core Ultra 7 256V processor.

browser.geekbench.com

Barely any improvement in ST

 

[511]

Yes it is, the ST is stagnant

It's nearly the same Copy Pasted Core what would you expect from a copy paste on a new node better perf/watt ?

It is in a sense that Elite X2 and Zen 6 is coming. One step at a time I guess.

Zen 6 will be competing with NVL

Pantherlake is Icelake 2025. CPU is meh, GPU is a big advancement, and the node is still immature. So 18A-P next year might be like Tigerlake and get oversized gains. The difference this time is that the iGPU was already competitive, when in Icelake it was not.

Ice lake and Panther Lake are like way apart Ice lake hadd too much clock regression vs 14nm

Intel simply followed Tik Tok strategy here to derisk node

 

[poke01]

Look here

Dell Inc. XPS 14 DA14260 - Geekbench

Benchmark results for a Dell Inc. XPS 14 DA14260 with an Intel Core Ultra X7 358H processor.

browser.geekbench.com

HP HP OmniBook Ultra Flip Laptop 14-fh0xxx - Geekbench

Benchmark results for a HP HP OmniBook Ultra Flip Laptop 14-fh0xxx with an Intel Core Ultra 7 256V processor.

browser.geekbench.com

Barely any improvement in ST

compared to Lunar Lake, Panther offers basically no improvements in ST.

That’s why you saw zero cpu benchmarks or tests based on CPU applications at CES.

Edit: Intel used speedometer 3.1 but only to say Panther lake uses less power than a prior CPU but didn’t actually show the performance improvements.

 

[DavidC1]

Ice lake and Panther Lake are like way apart Ice lake hadd too much clock regression vs 14nm

Intel simply followed Tik Tok strategy here to derisk node

It's same thing. Pantherlake's clock gap is much less compared to Arrowlake, but so is the uarch differences. Icelake and Skylake had big uarch differences but so did clock. Icelake gained nothing in CPU performance. Processes of both chips were immature, and both iGPUs were great gains over the predecessor.

It's nearly the same Copy Pasted Core what would you expect from a copy paste on a new node better perf/watt ?

Did you actually see the presentation? It's like a Tick+.

So for long term Intel, the question is this again. Will the same "rhyme" with 18A-P being Tigerlake, and we'll have a big gain a year after that but decline again? Intel doesn't seem to be out of the woods yet.

This is just like looking at graphs of any investments. In a bear market, it never declines straight down. In a bull market, it never rises straight up. Intel still is in a long term "bear" trend. Pantherlake is just a rising blip in the long term decline, maybe.