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.

 

[Kocicak]

... Intel has stated that they were too agressive with 10nm and tried to do too many things at once and it bit them.........
....
Intel is now saying that Meteor Lake will include 1) Tiles, 2) A die shrink, 3) Movement to EUV, and 4) A new architecture.

Anyone but me think that is an awful lot of risk for a single product to be exposed to?
...
My last concern is that Intel is laying off engineers. Seems hard to see how they are going to do 3 times the work with less people.

One thing is researching and developing a new thing and another putting things, which are already well researched and tested, together.

I believe that laying off employees which the company does not need is good for the company and for the employees as well, some of them were surely underutilised and not developing their skills and growing at all.

 

[Glo.]

Okay. Well, rumour had it for years that MTL might be Mobile only. The new roadmap seems to support that.
So my current expectation is aligned to many others here that MTL will start Mobile only and might go to the Mainstream DT segment later on (same compute tile), while ARL will take the High-End.

IMO, considering that for 2023 we will have Raptor Lake refresh, we are looking at mobile lineup, only for MTL-P, and Arrow Lake will happen for desktop.

Obviously. MTL-P and Arrow Lake-P also will happen on desktops in the form of NUCs.

But I wouldn't exclude one, another possibility of MTL-P and ARL-P happening on desktops - coming soldered to standard mITX, mATX boards, with ability to add your own hardware: (RAM, cooling, SSD, etc).

 

[Hulk]

Okay. Well, rumour had it for years that MTL might be Mobile only. The new roadmap seems to support that.
So my current expectation is aligned to many others here that MTL will start Mobile only and might go to the Mainstream DT segment later on (same compute tile), while ARL will take the High-End.

This makes sense. As I posted before I believe Meteor Lake is going to be the new Broadwell. Intel needs time to get the node going. On the positive side the transistor density will be higher and the cores improved. But they aren't going to be able to hit current Raptor Lake frequencies much less Raptor Refresh frequencies.

Intel doesn't want to release next generation part that is obviously less performant than the previous generation. Since Meteor will be more efficient than Raptor Lake the workaround is to release it only for mobile to allow them some time to add more cores, increase frequency, and IPC so that the next gen is better than Raptor on the desktop.

All of this cranking the clocks to 6GHz+ in Raptor and 24 cores to compete with AMD has the backlash of setting a high bar for them as well for the next gen on the desktop!

 

[mikk]

Intel accidentally releases test driver for DG3 "Elasti" dGPU, Raptor/Meteor/Arrow/Lunar-Lake iGPUs - VideoCardz.com

Intel accidentally provides support for (yet) non-existing architectures A graphics test driver that was not supposed to be published on a public website is now available on the Intel website, as spotted by leaker KOMACHI. Intel NUC Kit 11th Gen driver 30.0.101.9999 a test driver with support...

videocardz.com

So MTL-P 192 EUs, ARL-P - 384 EUs.

Apparently mobile MTL is limited to 128 EUs: https://www.igorslab.de/en/intel-me...e-block-diagram-of-mobile-14-generation-leak/

Because of the low clock speed on current Xe iGPUs (and poor low power CPU efficiency) we should see a big improvement nevertheless, the low clock speeds are fixed on Xe HPG+TSMC.

 

[eek2121]

So hold on for a moment, guys.

Meteor Lake and Arrow Lake mobile are coming 2023 and 2024 respectively?

🤣

So basically, nothing has changed except Intel refreshing Raptor Lake on the desktop.

Meteor Lake likely to see a limited mobile 2023 release along with Raptor Lake Refresh filling in the rest of 14th gen.

Arrow Lake, possibly Meteor Lake, and also possibly Raptor Lake (for lower SKU, high volume parts) will make up the 15th gen in 2024.

 

[Glo.]

Apparently mobile MTL is limited to 128 EUs: https://www.igorslab.de/en/intel-me...e-block-diagram-of-mobile-14-generation-leak/

Because of the low clock speed on current Xe iGPUs (and poor low power CPU efficiency) we should see a big improvement nevertheless, the low clock speeds are fixed on Xe HPG+TSMC.

It makes perfect sense now.

MTL-U 4P/8E, 128 EU
MTL-P/H 6P/8E, 128 EU.
MTL-P, 6P/16E/192 EU
ARL-P 8P/16E, 384 EU.

Within MTL-P there should be two designs. 6P/8E and 6/16E. IMO, if 4P/8E design exists, for MTL-U, then the lower end MTL-P will simply use iGPU die from that design.

NUCs with those SOCs will be epic.

 

[mikk]

A Raptor Lake refresh in H2 2023 means MTL-S won't come early 2024 I think.

It makes perfect sense now.

MTL-U 4P/8E, 128 EU
MTL-P/H 6P/8E, 128 EU.
MTL-P, 6P/16E/192 EU
ARL-P 8P/16E, 384 EU.

Within MTL-P there should be two designs. 6P/8E and 6/16E. IMO, if 4P/8E design exists, for MTL-U, then the lower end MTL-P will simply use iGPU die from that design.

NUCs with those SOCs will be epic.

Click the link, MTL mobile goes up to 128 EUs.

 

[Glo.]

A Raptor Lake refresh in H2 2023 means MTL-S won't come early 2024 I think.



Click the link, MTL mobile goes up to 128 EUs.

Which die design?

You realize that Igor's info is talking about 6P/8E design, whereas we have another MTL-P design leaked with 6P/16E config?

And that drivers info suggests that MTL-P tops at 192 EUs?

 

[mikk]

Which die design?

You realize that Igor's info is talking about 6P/8E design, whereas we have another MTL-P design leaked with 6P/16E config?

And that drivers info suggests that MTL-P tops at 192 EUs?

The biggest variant. I'm not talking about Igor, I'm talking about the Intel slide. And mobile MTL is limited to 6+8 not 6+16.

 

[mikk]

Meteor Lake-P LPDDR5x-6400/LPDDR5x-7500
Arrow Lake-PX LPDDR5x-8400
Lunar Lake-MX LPDDR5x-8500

https://twitter.com/x/status/1601656057885265921

 

[Hulk]

Unsaid agreement between AMD and Intel.

"Let's just cool it with the desktop rivalry for a bit and make some money okay?"

I think the 3D versions of Zen 4 aren't going to show the level of improvement that Zen 3 did because of the faster memory subsystem of Zen 4. Clocks may also come down, which could make them a niche gaming improvement like Zen 3 3D. So it'll be Zen 4 3D vs. Raptor refresh. If Zen 5 comes before Arrow, which is likely the case then that should put AMD in the driver's seat again.

 

[Glo.]

LPDDR5X - soldered to motherboards.

 

[Exist50]

LPDDR5X - soldered to motherboards.

Or CAMM, at least in theory.

 

[eek2121]

Unsaid agreement between AMD and Intel.

"Let's just cool it with the desktop rivalry for a bit and make some money okay?"

I think the 3D versions of Zen 4 aren't going to show the level of improvement that Zen 3 did because of the faster memory subsystem of Zen 4. Clocks may also come down, which could make them a niche gaming improvement like Zen 3 3D. So it'll be Zen 4 3D vs. Raptor refresh. If Zen 5 comes before Arrow, which is likely the case then that should put AMD in the driver's seat again.

Clocks aren't coming down on the Zen 4 variants. There is no agreement, why do you think AMD had to lower prices? Intel launched at a lower price. Zen3d products based on 5nm have been known to be coming for a while now, both via rumors and official slides from AMD. We've known Intel's roadmap for literally months, actually, possibly years IIRC. It hasn't changed. Certain individuals have tried to claim node delays, chip delays, etc. yet despite all that, the roadmap is the exact same as it was the first time we saw it with the exception of RPL-S refresh being added to an unknown spot that was previously empty. MTL-S hasn't been pushed out, and neither has Intel 4, 3, 18a, etc. Actually, as was pointed out early, some things are coming out ahead of schedule. This time next year we will see an RPL-S refresh. Nobody knows what that will consist of. Intel will launch a new product stack after spending a few months trailing behind AMD's X3D chips. AMD will eventually follow up, and we'll see the whole Coffee/Rocket/Alder/Raptor/Zen 3/Zen3d/Zen4 nonsense play out again practically verbatim.

If Intel/AMD were going to have a gentlemen's agreement it would be on price, not on product launches. You don't spend millions in R&D and not launch because your competitor asks you to, indirectly or otherwise.

 

[Hulk]

Clocks aren't coming down on the Zen 4 variants. There is no agreement, why do you think AMD had to lower prices? Intel launched at a lower price. Zen3d products based on 5nm have been known to be coming for a while now, both via rumors and official slides from AMD. We've known Intel's roadmap for literally months, actually, possibly years IIRC. It hasn't changed. Certain individuals have tried to claim node delays, chip delays, etc. yet despite all that, the roadmap is the exact same as it was the first time we saw it with the exception of RPL-S refresh being added to an unknown spot that was previously empty. MTL-S hasn't been pushed out, and neither has Intel 4, 3, 18a, etc. Actually, as was pointed out early, some things are coming out ahead of schedule. This time next year we will see an RPL-S refresh. Nobody knows what that will consist of. Intel will launch a new product stack after spending a few months trailing behind AMD's X3D chips. AMD will eventually follow up, and we'll see the whole Coffee/Rocket/Alder/Raptor/Zen 3/Zen3d/Zen4 nonsense play out again practically verbatim.

If Intel/AMD were going to have a gentlemen's agreement it would be on price, not on product launches. You don't spend millions in R&D and not launch because your competitor asks you to, indirectly or otherwise.

My comment was meant to be tongue and cheek humor with a hint of reality behind it. If Zen 3 had a longer run before Alder came along AMD would have really cleaned up. And vice-versa if Intel had got Alder out first. Then they both launched their next gens at the same time, sparking yet another price war. Just imagine if Raptor wasn't available until summer '23. AMD could have held prices and sold every part TMSC could manufacture for them.

AMD and Intel have been savaging each other and to our benefit! That's all I was pointing out.

 

[DrMrLordX]

I think the 3D versions of Zen 4 aren't going to show the level of improvement that Zen 3 did because of the faster memory subsystem of Zen 4.

L3 is still substantially lower in latency than system RAM on Zen4. Improvements should be quite large.

 

[Hulk]

L3 is still substantially lower in latency than system RAM on Zen4. Improvements should be quite large.

I forgot to mention the other part of my reasoning is that Zen 4 has significantly improved cache structure as compared to Zen 3, as well as a doubled in size L2.
Hopefully I'm wrong and Zen 4 3D has the same clocks as current parts and much improved performance across the board.

 

[LightningZ71]

Meteor Lake-P LPDDR5x-6400/LPDDR5x-7500
Arrow Lake-PX LPDDR5x-8400
Lunar Lake-MX LPDDR5x-8500

https://twitter.com/x/status/1601656057885265921

They intend on keeping 384EU of Xe fed with 128 bits of LPDDR5X 8400? Something doesn't math right...

 

[Glo.]

They intend on keeping 384EU of Xe fed with 128 bits of LPDDR5X 8400? Something doesn't math right...

128 bit 8400 MHz LPDDR5X should be around 134 GB/s if my math is correct.

But overall yeah, it may be a little too little. So here comes Intel ADM tech, that was touted for Arrow Lake.

 

[coercitiv]

I forgot to mention the other part of my reasoning is that Zen 4 has significantly improved cache structure as compared to Zen 3, as well as a doubled in size L2.
Hopefully I'm wrong and Zen 4 3D has the same clocks as current parts and much improved performance across the board.

I don't think you realize just how big of an advantage the 3D cache added to Zen 3. New gen CPUs run 20-25% faster clocks with fast DDR5 and pumped up cache structure only to achieve 5-11% better gaming performance. Cache size is clearly more important in many games.

 

[LightningZ71]

128 bit 8400 MHz LPDDR5X should be around 134 GB/s if my math is correct.

But overall yeah, it may be a little too little. So here comes Intel ADM tech, that was touted for Arrow Lake.

They could be intending to run it very slowly to keep it power efficient. That's a bold strategy cotton...

 

[Hulk]

I don't think you realize just how big of an advantage the 3D cache added to Zen 3. New gen CPUs run 20-25% faster clocks with fast DDR5 and pumped up cache structure only to achieve 5-11% better gaming performance. Cache size is clearly more important in many games.

You are probably right. Thing is 5800X3D results are all over the map for games. In some games huge benefit and others not so much. At higher resolutions current CPUs can push really high frame rates so I'm not sure of the benefit.

In not-gaming applications benefits are even more inconsistent. Also for some reason clock is 400MHz lower than the normal version. We shall see. It is still my belief that Zen 3, being super wide was not properly "fed" instructions and this limitation was largely mitigated with Zen 4 and the improved caching structures and larger L2, along with DDR5 memory.

As I wrote before I'll be happily surprised to be wrong. Admittedly I didn't think Zen 4 would show so much improvement over Zen 3. The IPC improvement, yes, saw that coming. But the increase in clocks was a total surprise for me. The AMD engineers are very on the ball so I'm sure they wouldn't do this if the results weren't there. I simply have doubts. We shall see.

 

[coercitiv]

Thing is 5800X3D results are all over the map for games.

Are they? Here's a graph for 5800X3D vs. 7600X and another one for 7600X vs. 13600K. If we were to mirror the second graph to have 7600X as the second comparison point, it would look quite similar in terms of overall distribution.

At higher resolutions current CPUs can push really high frame rates so I'm not sure of the benefit.

That is true for all value and midrange CPUs making high-end SKUs look bad in real gaming conditions.

Also for some reason clock is 400MHz lower than the normal version.

The reason was a shared voltage rail for core and cache. That's been addressed in Zen 4, cache has it's own voltage plane.

 

[Hulk]

Are they? Here's a graph for 5800X3D vs. 7600X and another one for 7600X vs. 13600K. If we were to mirror the second graph to have 7600X as the second comparison point, it would look quite similar in terms of overall distribution.

View attachment 72739


That is true for all value and midrange CPUs making high-end SKUs look bad in real gaming conditions.


The reason was a shared voltage rail for core and cache. That's been addressed in Zen 4, cache has it's own voltage plane.

Well that is your interpretation. I tend to agree with Gavin Bonshor in his conclusion of his review of the 5800X3D.

"Looking at our data across our CPU testing suites (compute and gaming), the AMD Ryzen 7 5800X3D is a mixed bag. "

Furthermore...
"While AMD pitched the Ryzen 7 5800X3D and its 96 MB of L3 cache as being beneficial in gaming, they didn't make the same kind of claims when it comes to applications and general compute performance. Because the benefits of a larger L3 cache are even more sporadic in these cases, AMD has wisely opted not to promote the part based on general compute performance here. Which is not to say that it's bad news for AMD, but as our data confirms, the L3 cache generally doesn't have much of an impact here. "

And finally...
"Even at $449, the Ryzen 7 5800X3D can perform as well, if not at times better than any chip from Intel's 12th Gen Core series of processors. The catch is that, as is often the case with new/unique technologies, the gains are uneven among games. This essentially means if the game doesn't benefit from the large 96 MB of 3D V-Cache onboard, it performs similar to the already existing Ryzen 7 5800X, which can be had for $350; $100 cheaper than the 5800X3D. In that respect, it would be helpful if AMD published a guide of games that they've found to materially benefit from the large L3 cache, just so gamers and users could decide whether the Ryzen 7 5800X3D is worth it. "

 

[ondma]

Are they? Here's a graph for 5800X3D vs. 7600X and another one for 7600X vs. 13600K. If we were to mirror the second graph to have 7600X as the second comparison point, it would look quite similar in terms of overall distribution.

View attachment 72739


That is true for all value and midrange CPUs making high-end SKUs look bad in real gaming conditions.


The reason was a shared voltage rail for core and cache. That's been addressed in Zen 4, cache has it's own voltage plane.

Those graphs are sort of deceiving because the scale has been condensed to magnify the differences. I pulled up the graphs from the HUB site itself and it looks like 5800x3D vs 7600x ranges from 15% faster to 35% slower. So the total variance is about 50%. That seems like "all over the place" to me.