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.

 

[Meteor Late]

If I'm right, Fire Range is enthusiast series targeted at gamers. Does AMD have any mainstream series like Intel's 285H, 265H, etc) that's targeted at general consumers?

I'm assuming HX370 is the consumer series. If thats the case, AMD's laptop parts this gen is a lost cause.

The lost cause is the price, not the performance.
This battle about who has the highest MT in a laptop is stupid from my POV.
Fire Range is even more stupid I agree, calling that a Laptop chip is just a joke.

 

[Markfw]

The lost cause is the price, not the performance.
This battle about who has the highest MT in a laptop is stupid from my POV.
Fire Range is even more stupid I agree, calling that a Laptop chip is just a joke.

I sort of agree. But there are those that use their laptop for everything, decoding and encoding, etc. For me I just need to do internet, as all the power I have is at home. So good ST and battery life and price are the most important.

 

[OneEng2]

The lost cause is the price, not the performance.
This battle about who has the highest MT in a laptop is stupid from my POV.
Fire Range is even more stupid I agree, calling that a Laptop chip is just a joke.

Interesting POV.

I have to wonder how many people still have desktops at home (I do of course). Windows (and pretty much all modern OS's) eat up a pretty good amount of threads all on their own, so I don't think that single core processors are going to be making a big comeback either.

I do wonder how many people do serious work (video editing, CAD) on a laptop? Software development for sure. I haven't seen a SW engineer using anything but a laptop for some time.

So why doesn't Intel or AMD make a super small, inexpensive 4 core processor for laptop? They could likely make 2 of those cores clock to the moon.

 

[Thunder 57]

Interesting POV.

I have to wonder how many people still have desktops at home (I do of course). Windows (and pretty much all modern OS's) eat up a pretty good amount of threads all on their own, so I don't think that single core processors are going to be making a big comeback either.

I do wonder how many people do serious work (video editing, CAD) on a laptop? Software development for sure. I haven't seen a SW engineer using anything but a laptop for some time.

So why doesn't Intel or AMD make a super small, inexpensive 4 core processor for laptop? They could likely make 2 of those cores clock to the moon.

For AMD it would mean disabling half the cores since the have 8 core CCX's. Intel might be able to but really how much higher do you expect cores to clock?

 

[igor_kavinski]

Interesting comparison: https://www.overclock.net/posts/29408377/

9800X3D's power and temps are SWEET.

Those high RAM speeds will help Intel sell Arrow Lake when CUDIMM becomes affordable.

 

[SiliconFly]

.

 

[511]

MT? Nope. Arrow Lake 285H beats AMD HX 370/375 in both ST & MT by a wide margin.

I think it will be 5-7% at best

Nope. Both Fire Range & ARL-HX are awesome and are meant for a very specific segment. Not for regular laptop users, but for gamers & enthusiasts who're looking for extreme perf and not bothered with battery life, heat, weight, etc. I prefer HX series (don't mind paying a little extra and also it's always connected to the power outlet). And they're always coupled with a powerful dGPU. Essentially a desktop replacement in a mobile form factor (but not so mobile like regular laptops due to weight, heat and relatively poor battery life).

AMD & Intel HX series typically tend to be blazing fast compared to regular laptop offerings from Intel, AMD, Apple, QC, etc. Imagine a 285 HX coupled with a RTX 5080... may hit a mind blowing >1000fps at max settings in Doom Eternal. Just saying.

AMDs fire range has the worst issue of the CPU Sucking big time in laptop due to their Chiplets so ARL-HX will simply get better standby and idle

 

[SiliconFly]

.

 

[511]

Big AVX10 Leak 🫠🫠

https://twitter.com/x/status/1873299846729470029

 

[Kepler_L2]

Big AVX10 Leak 🫠🫠

https://twitter.com/x/status/1873299846729470029

Old news

 

[Meteor Late]

Not sure about fire range. But I remember reading their last gen equivalent was real good. So it should translate well into this gen Fire Range also.

Idle power was garbage. While I believe Strix Halo won't be that despite using chiplets, because it seems it is different, Fire Range seems exactly the same as Dragon Range.

 

[poke01]

Idle power was garbage. While I believe Strix Halo won't be that despite using chiplets, because it seems it is different, Fire Range seems exactly the same as Dragon Range.

Halo should be much lower fire range.

 

[511]

Old news

At least it's official from Intel principal engineer 🫠🫠

 

[SiliconFly]

.

 

[OneEng2]

Does anyone have a guess about the difference in performance between AVX512 full vs. double pumped AVX256?

Where do we expect Intels AVX10 to be performance wise in comparison?

BTW, I hate Intel's constant moving of the goal post by introducing new instructions and getting benchmarks to support them prior to launch.

Why not just put AVX512 data path back?

Seems like Intel is still utilizing their monopoly power to rig the game. I suppose that still works when the market is 70/30. By the time AVX10 Rolls out it may look more like 65/35. Should still work though.

 

[techjunkie123]

Does anyone have a guess about the difference in performance between AVX512 full vs. double pumped AVX256?

Where do we expect Intels AVX10 to be performance wise in comparison?

BTW, I hate Intel's constant moving of the goal post by introducing new instructions and getting benchmarks to support them prior to launch.

Why not just put AVX512 data path back?

Seems like Intel is still utilizing their monopoly power to rig the game. I suppose that still works when the market is 70/30. By the time AVX10 Rolls out it may look more like 65/35. Should still work though.

Phoronix has a piece on this: https://www.phoronix.com/review/amd-epyc-9755-avx512

256double pump works well enough for many applications.

 

[SiliconFly]

.

 

[511]

Does anyone have a guess about the difference in performance between AVX512 full vs. double pumped AVX256?

Where do we expect Intels AVX10 to be performance wise in comparison?

BTW, I hate Intel's constant moving of the goal post by introducing new instructions and getting benchmarks to support them prior to launch.

You don't create instructions in such a complicated ISA for no reason there must be a reason they are created good Examples are VNNI/AES/SHA bad ones are VPIntersect

Why not just put AVX512 data path back?

Seems like Intel is still utilizing their monopoly power to rig the game. I suppose that still works when the market is 70/30. By the time AVX10 Rolls out it may look more like 65/35. Should still work though.

The problem is AMD doens't have the resources to put into building the Software ecosystem for x86 it is still relying on Intel to do the gruntwork it is the reason they are failing in GPU race cause they can't magically run CUDA(zluda exists) like they can with x86 and AVX10 is a good step from point of view of many SW people the only issue is the data path

 

[Hitman928]

Not sure. But if I'm right, Intel did have AVX-512 in clients at one point of time and it sucked. Something related to performance issues due to throttling or frequency issue or something. They might have decided to start with a clean slate instead with AVX10 for clients and AVX-512 for servers. And AVX10 should take up less die space than AVX-512 which is very important in clients.

And I don't think Intel has lost that much x86 market share in clients recently. It still has around 80% to 85% market share in clients and around 70% market share is servers.

View attachment 113943

Source: wccftech

(Note: Intel did lose some client market share at one point, but also regained the lost market share at a later time. In servers, they're still bleeding.)

Your numbers are a bit out of date.



source

 

[LightningZ71]

Intel's last two client products OFFICIALLY supporting AVX-512 were tiger lake and Rocket Lake. Rocket Lake was heavily thermally limited due to being back ported to 14++++. Tiger Lake was on 10sf, which was OK but still had power/thermal issues. I think that both implementations had half the throughput of the Xeon class units but were at least logically complete.

 

[SiliconFly]

.

 

[igor_kavinski]

Why not just put AVX512 data path back?

Because they aren't as inventive and good as AMD engineers: http://www.numberworld.org/blogs/2024_8_7_zen5_avx512_teardown/

With this, AMD has finally surpassed Intel in nearly every category related to SIMD execution - crushing them in 512-bit performance. This is a massive turn-around from the days of Bulldozer and Zen1. Intel has historically been the pioneer of SIMD. But now, AMD has taken the crown from them and beaten them at their own game.

Rather than be compared to AMD's stellar AVX-512 implementation, they simply decided to pull out of the race to avoid unnecessary embarrassment in EVERY review.

 

[511]

Because they aren't as inventive and good as AMD engineers: http://www.numberworld.org/blogs/2024_8_7_zen5_avx512_teardown/

LOL really? AVX-512 won't exist if Intel was not inventive i would give AMD Credit for their AVX-512 Implementation but this is a bit too far

Rather than be compared to AMD's stellar AVX-512 implementation, they simply decided to pull out of the race to avoid unnecessary embarrassment in EVERY review.

They threw the towel cause their P/E core dilema

 

[SiliconFly]

.

 

[igor_kavinski]

They threw the towel cause their P/E core dilema

If they had been so good, they could've put in the extra effort to keep AVX-512. Haven't heard of Zen4c/5c?