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.

 

[DavidC1]

Halo is N4P + tech from 2022/2023 (by 90%, the rest is LP, unified memory etc)

And what's your point? Intel in 2019 vs Zen 3: "Intel has an outdated core from 2015 with 14nm process versus AMD's 2019 core and 7nm process". Were you defending Bulldozer in 2010?

AMD chose not to implement the latest tech in their consumer lineup.

Battery life might be pretty good on Panther Lake. That's about it. Intel shouldn't be charging THAT MUCH just for some extra battery life though.

They are not though. Pantherlake is only 30 Euros(1770) versus more than AMD's Gorgon Point(1740). If Pantherlake goes up in price compared to predecessor, so does AMD's Gorgon Point over Strix Point.

@Hulk Integer differences can't be explained by memory system differences. It's mostly uarch. SpecFP at some point is basically a memory benchmark. Also Geekerwan shows different results for E vs LPE.

It's because Zen 5C is clocked lower, and ipc generally decreases as clock speed increases, esp if u are memory bound.

Ding Ding Ding! Likely in that review they just took the score and divided by the GHz numbers. In Geekerwan's review, it looks like they actually took the liberty of lowering the P core clocks.

 

[DavidC1]

The 4 Xe core version doing that well reminds me when they introed Celeron D based on the Prescott core and that chip beat the Northwood based Celeron by 25%+. The uarch gains are having oversized gains in lower end where it's artificially segmented, thus crippled. It only has 4MB cache, meaning it's uarch doing all the work.

Also, this is actually proof that it can't be memory bandwidth bound. Actually there's two proofs. One is that with 1.5x compute it gets 1.7-1.8x gains. And the other is the 4 Xe core version with only 4MB cache is doing this well.

Wildcatlake's 2 Xe can probably beat 96 EU in Iris Xe of pre-Meteorlake processors. That means ~4x faster than predecessor, allowing low setting gaming for modern titles.

 

[fastandfurious6]

And what's your point? Intel in 2019 vs Zen 3: "Intel has an outdated core from 2015 with 14nm process versus AMD's 2019 core and 7nm process". Were you defending Bulldozer in 2010?

One big difference among many:

In 2019, Intel didn't have any better tech to implement 😉 they waited for Alderlake which saved them.

AMD does have plenty: Vcache, RDNA4... isn't 2nm Zen6 also almost ready?

 

[DavidC1]

AMD does have plenty: Vcache, RDNA4... isn't 2nm Zen6 also almost ready?

It makes no difference. It's all a black box. None of what we argue here makes an ounce of difference. AMD chose not to implement them. Right now Intel is the more consumer friendly vendor. They actually talked decently about PCs, versus AIMD and NVAIDIA.

It seems you always need to feed people a slice of humble pie once in a while at least.

 

[Hulk]

Looking back the ST IPC equivalence of new/old cpu's is astounding.
Cinebench accelerators/E core spam used to be our top dog Skylake
Then the spam morphed into Golden/Raptor Cove!
Are they still spam I wonder?

Skylake~Gracemont~Zen 3
Raptor/Golden Cove~Skymont~Zen 4
Lion Cove~Zen 5

 

[Tachyonism]

MSI Prestige with Panther Lake 385h is $1300.

Do you mean the X7 358H instead?

 

[511]

Looking back the ST IPC equivalence of new/old cpu's is astounding.
Cinebench accelerators/E core spam used to be our top dog Skylake
Then the spam morphed into Golden/Raptor Cove!
Are they still spam I wonder?

I wouldn't count Multiple Cores with Int/FP of Golden Cove Spam cores.

Skylake~Gracemont~Zen 3
Raptor/Golden Cove~Skymont~Zen 4
Lion Cove~Zen 5

Zen 3 is a decent bit better than Skylake gracemont it's in the same class as sunny cove iirc

 

[OldGamersNeverDie]

Do you mean the X7 358H instead?

I did, my bad.

 

[Hulk]

I wouldn't count Multiple Cores with Int/FP of Golden Cove Spam cores.

Zen 3 is a decent bit better than Skylake gracemont it's in the same class as sunny cove iirc

I don't think they are spam either. But some in here do.

For ST, Gracemont and Skylake are pretty even, but not core-for-core.

 

[DavidC1]

I don't think they are spam either. But some in here do.

For ST, Gracemont and Skylake are pretty even, but not core-for-core.

It seems Gracemont has the capability to do it, but in actual implementations held back by the shared L2 cache and generally slower uncore.

But Skymont shows that it was close enough that you can substitute instead of the P core in even gaming scenarios. Remember the Arrowlake test where 1+16 is faster than 8+16? Or how in Pantherlake they are prioritizing E instead of P?

Jim Keller said Skymont was made to directly compete with Zen 5. With Gracemont it was architecturally a slower core than Zen 4C. It no longer is the case.

 

[DrMrLordX]

Wasn’t that intended/expected and 18AP and 14A will be for higher clocking CPUs?

Diamond Rapids-SP was supposed to be 18a. Diamond Rapids-AP may wind up being on 18AP.

Was there time and effort for 18a that can’t be used for upcoming nodes and products?

Sorry, but what?

They are not though. Pantherlake is only 30 Euros(1770) versus more than AMD's Gorgon Point(1740).

Where are the sub-$1000 Pantherlakes? This is supposed to be a low-power SoC that doesn't even operate above 25W. It's like Lunar Lake all over again. Also how is Gorgon Point relevant when Pantherlake is competing with the now-quite-old Strix Point?

 

[regen1]

Diamond Rapids-SP was supposed to be 18a. Diamond Rapids-AP may wind up being on 18AP.

No. DMR-SP was 18A-P version same as DMR-AP.

 

[Josh128]

Re: The Strix Halo vs Panther Lake pricing debate, I will say that at least for gaming, SKUs like below blow both of them out of the water for the money. This is a real time availability ad from Best Buy US. PL and Halo both need to be available at significantly sub $2k to even be taken seriously in the gaming conversation. Also, from Amazon, it appears that available Halo laptops are basically non-existent, so theres that too. Halo is indeed a non-competitor unless AMD does something that drastically changes availability.

 

[Magio]

Halo is indeed a non-competitor unless AMD does something that drastically changes availability.

View attachment 137624

Well, them launching more varied Halo SKUs at CES may improve the situation somewhat but I think it goes down to Halo being in a very awkward place where you probably just want an Nvidia dGPU if you're paying that much for silicon and operating at the TDP levels where Halo shines. Even in the edge cases where Halo would actually be the more efficient option it would still be in a segment where Nvidia's brand appeal just wins.

 

[511]

Re: The Strix Halo vs Panther Lake pricing debate, I will say that at least for gaming, SKUs like below blow both of them out of the water for the money. This is a real time availability ad from Best Buy US. PL and Halo both need to be available at significantly sub $2k to even be taken seriously in the gaming conversation. Also, from Amazon, it appears that available Halo laptops are basically non-existent, so theres that too. Halo is indeed a non-competitor unless AMD does something that drastically changes availability.

View attachment 137624

you can get this one as well it's 64GB ram for extra 330$



Why are we arguing price of platform like Panther Lake that haven't even launched for 1 Month? New Platforms are expensive we have seen that with desktop/laptop what was the price of CPU when Zen 4 came out or ARL launched they were expensive

 

[OneEng2]

PTL has higher IPC than Zen5/5c:

https://wccftech.com/intel-panther-lake-powered-by-the-new-cougar-cove-p-cores-and-darkmont-e-cores-takes-a-lead-over-amd/

View attachment 137542

I do wonder though how valid these spec numbers are for everyday use? Seems like benchmarks like CB24 would indicate a significant lead for ARL yet in a suite of actual application benchmarks, Zen 5 wins handily (CB is a higher level bench than spec. I do get it, but the general point is where I am going here).

ARL-H is a terrible product, slower than zen5 in everything. How does it have higher IPC than Zen5 😂

... and this is where the rubber hits the road. I am looking forward to the benchmarks of applications.

Would like to see some results/tests from others but from these, PTL's cores got significant upgrade for a tick/refresh thing, some from optimized fabric/L3 and some from improvements to the cores.

... and this may become doubly true for NVL where the N2 process may provide some clock headroom for scaling GHz.

I always thought that ARL was severely hobbled by its latency issues with the ring bus. I figured that just like Zen 2, Intel's first try at tiles left them with some choke points they hadn't anticipated. Freeing up these issues for NVL/PTL could free up significant improvement along the same lines as the Zen 2 to Zen 3 move did.