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.

 

[Saylick]

area efficiency is down like 50% gen on gen.

That's just a natural consequence of the square root scaling law...

 

[igor_kavinski]

A true monster.

If it gets fed properly by the available RAM bandwidth and there are no serious latency issues. Can't say I'm optimistic after the sadness that is ARL.

 

[Dave3000]

Does anyone here know if a Ryzen 7 9800X3D beats a Core Ultra 9 285K in X-Plane 12?

 

[dullard]

And I struggle to see how Intel plans to double core counts at any reasonable TDP.

Easy: less power to each core. There are always have 35 W base and ~112 W turbo chips in Intel's top segment (formerly i9). There is absolutely no reason that you have to run Intel chips at high power. The reviewers just choose to feed them with massive power supplies, huge coolers, and unlimited power settings on the motherboard. But, they could just have easily ran the chips with less power. Performance would be different, but it is quite functional.

NVL-SK: 2x 8+16
NVL-HX: 1x 8+16
NVL-S / NVL-H: 4+8
NVL-U: 4+0

Videocardz is attributing that to Exist50 on Reddit. Is that our very own Exist50? https://videocardz.com/newz/intel-nova-lake-s-for-desktops-rumored-to-feature-2x8p16e-configuration

 

[Abwx]

Does anyone here know if a Ryzen 7 9800X3D beats a Core Ultra 9 285K in X-Plane 12?

At this site they know, with Windows updates comparisons for both CPUs.

Test du Ryzen 7 9800X3D : le nouveau maître du jeu !

Loin d'être le succès espéré, les Ryzen 9000 retrouvent de l'intérêt via une baisse de prix et les contreperformances ludiques des Core Ultra 200S. Le moment idéal pour enfoncer le clou avec une version à base de 3D V-Cache, débarrassée des contraintes thermiques limitant ses prédécesseurs.

hardwareand.co

 

[Hulk]

Easy: less power to each core. There are always have 35 W base and ~112 W turbo chips in Intel's top segment (formerly i9). There is absolutely no reason that you have to run Intel chips at high power. The reviewers just choose to feed them with massive power supplies, huge coolers, and unlimited power settings on the motherboard. But, they could just have easily ran the chips with less power. Performance would be different, but it is quite functional.


Videocardz is attributing that to Exist50 on Reddit. Is that our very own Exist50? https://videocardz.com/newz/intel-nova-lake-s-for-desktops-rumored-to-feature-2x8p16e-configuration

Exactly. v/f curve is nonlinear. Relatively small decrease in frequency provides a significant drop in power.

100 cores running at 5GHz don't use twice the power of 100 cores running at 2.5GHz. You could probably run 150 cores at 2.5GHz at that power envelope. Therefore in a well threaded application you see up to 50% increase in performance. The tradeoff of course if more die area (or more transistors on a smaller node). For a long while Intel "worked around" the die area issue by ramping up clocks. Eventually it caught up to them.

As I've written before I'd be happy with a Zen 6 that is 5% better IPC and 24 cores. Very happy actually. For my day-to-day usage I'm more MT bottlenecked than ST bottlenecked.
I run a few apps that are well threaded and I often run them simultaneously.
Like converting RAW photos in PureRaw4 while rendering out a video in Vegas Pro while editing in PS while working on a word doc. I need my PC to remain responsive and get stuff done. That requires a lot of cores. I max out my 9950X all the time.

 

[Abwx]

Exactly. v/f curve is nonlinear. Relatively small decrease in frequency provides a significant drop in power.

100 cores running at 5GHz don't use twice the power of 100 cores running at 2.5GHz.

100 cores at 5GHz use as much power as 400 cores at 2.5GHz, and it s even as much as 500 cores at 2.5Ghz given that around 5GHz power increase as a cube of frequency or so.

 

[DavidC1]

That's just a natural consequence of the square root scaling law...

He's just wrong. It's 30% area increase with 30% Int improvement + 20-30% area on top of that for doubled FP block, which got us 70% improvement.

Unless they are planning to double the FP block again, it's likely it'll be 1:1 increase again. Eg, 30% Int/FP for 30% area.

100 cores at 5GHz use as much power as 400 cores at 2.5GHz, and it s even as much as 500 cores at 2.5Ghz given that around 5GHz power increase as a cube of frequency or so.

This is based on the nonsensical assumption we're still in the late 90's, early 2000's where there was PLENTY of voltage scaling. It stopped once it reached about ~1.5V.

There's nothing easy anymore. NOTHING. It's obvious to anyone not lying to themselves and have played with simple GPU voltage/frequency scaling. At about 0.6V it can barely reach 300MHz.

From that point it scales superlinearly in regards to voltage - 5-10% voltage may double frequency, or 1.05x1.05 x 2 = 2.2x power for 2.2x frequency assuming we ignore static leakage+fixed block power.

Exactly. v/f curve is nonlinear. Relatively small decrease in frequency provides a significant drop in power.

V/F curve has a superlinear, linear, and sublinear relation depending on what frequency curve you are at. It's not that simple.

It used to be simple as @Abwx claims back in the 0.35u days when the voltages were in the 2-2.5V range. Not anymore. The transistors in modern CPUs have been stuck not far above threshold voltage for a long, long time.

 

[desrever]

He's just wrong. It's 30% area increase with 30% Int improvement + 20-30% area on top of that for doubled FP block, which got us 70% improvement.

So you think 30% area increase on 3NM vs 10NM isn't going to affect the calculations. The density increase from process alone would be >100%.

 

[Kepler_L2]

Unless they are planning to double the FP block again, it's likely it'll be 1:1 increase again. Eg, 30% Int/FP for 30% area.

They are moving to FP-256 to support AVX10

 

[511]

area efficiency is down like 50% gen on gen.

How against crestmont?

 

[511]

They are moving to FP-256 to support AVX10

Also APX and FRED
wanna leak something to make this thread more interesting?

 

[DavidC1]

So you think 30% area increase on 3NM vs 10NM isn't going to affect the calculations. The density increase from process alone would be >100%.

It's ISO-process, but let your bias cloud your judgment.

 

[Hulk]

He's just wrong. It's 30% area increase with 30% Int improvement + 20-30% area on top of that for doubled FP block, which got us 70% improvement.

Unless they are planning to double the FP block again, it's likely it'll be 1:1 increase again. Eg, 30% Int/FP for 30% area.

This is based on the nonsensical assumption we're still in the late 90's, early 2000's where there was PLENTY of voltage scaling. It stopped once it reached about ~1.5V.

There's nothing easy anymore. NOTHING. It's obvious to anyone not lying to themselves and have played with simple GPU voltage/frequency scaling. At about 0.6V it can barely reach 300MHz.

From that point it scales superlinearly in regards to voltage - 5-10% voltage may double frequency, or 1.05x1.05 x 2 = 2.2x power for 2.2x frequency assuming we ignore static leakage+fixed block power.

V/F curve has a superlinear, linear, and sublinear relation depending on what frequency curve you are at. It's not that simple.

It used to be simple as @Abwx claims back in the 0.35u days when the voltages were in the 2-2.5V range. Not anymore. The transistors in modern CPUs have been stuck not far above threshold voltage for a long, long time.

I was referring to the part of the curve that is generally relevant to our discussions, which is of course at the higher voltages and frequencies where things are not linear at all.

 

[DavidC1]

They are moving to FP-256 to support AVX10

I still think Intel was retarded in pushing 512-bit FP on a CPU but since they did it anyway, if they straight up double the block again, it'll have superior peak throughput in 256-bit FP operations versus Zen 5 and definitely Lion Cove.

 

[511]

I still think Intel was retarded in pushing 512-bit FP on a CPU but since they did it anyway, if they straight up double the block again, it'll have superior peak throughput in 256-bit FP operations versus Zen 5 and definitely Lion Cove.

There are lots of retard at the company 😂

 

[desrever]

It's ISO-process, but let your bias cloud your judgment.

If you say so lol.

Skymont performance improvement barely makes up for losing HT from RPT vs ARL. This is going from less than 1/3 the size of the respective P core to almost 1/2 the size of the P core.

Shrinking Raptor lake to 3nm would probably yeild more area efficient cores than either of the ones in Arrow lake.

Especially since people keep talking core area, not including L2 cache, which means logic scaling will dominate. If we assume Intel 7 roughtly equals N7 in logic density then:

With 3x the logic density, it would have to perform a lot better to keep the same perf/area ISO process.

 

[511]

If you say so lol.

Skymont performance improvement barely makes up for losing HT from RPT vs ARL. This is going from less than 1/3 the size of the respective P core to almost 1/2 the size of the P core.

Shrinking Raptor lake to 3nm would probably yeild more area efficient cores than either of the ones in Arrow lake.

Especially since people keep talking core area, not including L2 cache, which means logic scaling will dominate. If we assume Intel 7 roughtly equals N7 in logic density then:

With 3x the logic density, it would have to perform a lot better to keep the same perf/area ISO process.

Uhh those are HD Libs btw and then 291mm2 number is based on 1-1 Fin library that no one uses at minimum designer use 2-1 and for CPU 3-2/3-3 is the most common library

 

[511]

Btw so we have leaked NVL Cores any idea how much ST increase will we get this much MT is too much for most of the people

 

[igor_kavinski]

Does anyone here know if a Ryzen 7 9800X3D beats a Core Ultra 9 285K in X-Plane 12?

Microsoft Flight Simulator 2024 (Asobo Studio) | So after some (very little) consideration I returned my 285K and Z890 and got myself a 9800X3D and X870E… and got even some money back 😆 | Facebook

So after some (very little) consideration I returned my 285K and Z890 and got myself a 9800X3D and X870E… and got even some money back 😆. Performance is now so much better and I am no longer so CPU...

www.facebook.com

 

[SiliconFly]

.

 

[Hulk]

There are some cases where the 285K is more performant than the 9950X but you have to look closely at those results. I noticed this when I was deciding between the 285K and 9950X.
For example, the 285K is faster when encoding x264, but not x265 or SV1.
x264 encoding speed isn't really relevant in my opinion for two reasons. x265 is the new standard and is more efficient with encoding and SV1 is next up.
In addition, x264 is not nearly as processor intensive as x265 so encoding speed really isn't an issue. Any processor from Skylake on can handle x264 encoding in a timely manner.

 

[SiliconFly]

.

 

[Hitman928]

285K media engine fully supports HEVC (H.265) hardware-accelerated encode/decode and should easily beat the competition. Which encoder are you using? Maybe the one you're using doesn't support/use Intel HEVC media engine. Also, Lunar Lake supports VVC (H.266).

Some people still prefer to encode with the CPU.

 

[Thibsie]

Some people still prefer to encode with the CPU.

And for very good reasons.