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.

 

[Io Magnesso]

If the feedback on this problem is a problem specific to MSI, It would be a problem if I sent it to Intel Intel has created XMP standards, but it's not making a motherboard. Better to report to the MSI forums.

 

[igor_kavinski]

Based on this, I have to rescind my praises for Arrow Lake and Z890 and sadly call it instable, I have never seen my event viewer throwing so many criticals and warnings withing an hour.
Arrow Lake gets a 2/10 after retrying the platform and wasting my time.

Have to question your weird conclusion. It's MSI's fault. Intel can't be held responsible for MSI botching their BIOS/mobo validation.

 

[AcrosTinus]

Have to question your weird conclusion. It's MSI's fault. Intel can't be held responsible for MSI botching their BIOS/mobo validation.

The weird conclusion comes from being mad about the wasted time, In my naive thought MSI and Intel are tight and the Intel SWEs would not letting something this bad pass. Literally crashing the system on a hourly basis. I will admit though that without insight, Intel's fault is minimal but still as a customer it feels bad. My first MSI board, what a let down.

 

[poke01]

The weird conclusion comes from being mad about the wasted time, In my naive thought MSI and Intel are tight and the Intel SWEs would not letting something this bad pass. Literally crashing the system on a hourly basis. I will admit though that without insight, Intel's fault is minimal but still as a customer it feels bad. My first MSI board, what a let down.

I would not buy MSI again. Thanks for investigating

 

[igor_kavinski]

My first MSI board, what a let down.

I haven't seriously considered an MSI mobo ever since I saw someone's failed SATA port corrupting HDD data back in the 845 chipset days. That was a brand new mobo not even 6 months old. Since then, I've had an MSI Geforce 1060 3GB which didn't crap out and I also have a MSI Z690 DDR4 mobo that I never found the time to use (idea was to flash modded BIOS to try to enable AVX-512 on my 12700K). My current rating for these Taiwanese manufacturers is ASROCK, Gigabyte, ASUS and MSI, in that order. ASUS is lower mainly because their prices are hardly ever attractive.

 

[Darkmont]

Intel’s Lion Cove P-Core and Gaming Workloads

Lion Cove is Intel’s latest high performance CPU architecture.

chipsandcheese.com

 

[DavidC1]

Why decode width will continue to scale:

When people give the argument that we don't need more decoders, the defense is that ILP is in average no more than 2, or 3.

However, the reason CPUs get wider is not for the average case. In United States, the average lifespan is about 80 years old. But we don't tell 80 year olds that "it's time for you to die", and you get no healthcare or any help. Average means there are cases where it's lower and where it's higher. You try to make the average higher by helping everyone, those that might die much earlier, those that might die at average, and those that are already past average.

Wider decoders exist for the peaks, because there are scenarios where ILP can go into many double digits.

A CPU doesn't only consist of decoders. It consists of many different blocks. Thus, from a power/area perspective, decoders are only a fraction of it. If for example a decoder block only takes 4% of the die, it would be worth doubling the amount if it nets you 4% extra performance.

Architecting a better general purpose CPU is basically finding what is the ideal balance between area and power for commonly used application base at the time.

Say you get 2% gain for each of the 5 basic blocks in the CPU: decoders, reservation stations and buffers, execution units, memory hierarchy, better branch prediction, then you end up with a respectable 10%. If you then only needed 2% area and power increase for each of the 5 basic blocks, then it's worth doing it.

If the supposed one block only took 2% of the area/power, then it's worth doubling that block, because while isolating it sounded like a bad tradeoff with 100% increase for just 2% performance, in the big picture you got 2% for extra 2% power/area.

 

[Io Magnesso]

chipsandcheese.com is Wasn't he disliked?

 

[Darkmont]

chipsandcheese.com is Wasn't he disliked?

By who, and why? Chips and Cheese does good work providing analysis at the microarchitectural and SOC design level.

 

[Io Magnesso]

By who, and why? Chips and Cheese does good work providing analysis at the microarchitectural and SOC design level.

No, there was a topic that I hated ChipsandCheese before.
I thought it was a bit contradictory
I am entertaining their content

 

[poke01]

Intel Panther Lake Xe3 Integrated Graphics Firmware Upstreamed - Phoronix

www.phoronix.com

firmware upstreamed

 

[511]

Intel Panther Lake Xe3 Integrated Graphics Firmware Upstreamed - Phoronix

www.phoronix.com

firmware upstreamed

It's launching In 2025 in limited quantity there is no doubt about it.

 

[Magio]

It's launching In 2025 in limited quantity there is no doubt about it.

Yeah if I had to guess I'd say late Q4 launch in very few of SKUs from Intel's top partners (Dell, Lenovo, Asus, HP?), followed by more models revealed at CES for release in Q1/Q2 2026.

 

[511]

Yeah if I had to guess I'd say late Q4 launch in very few of SKUs from Intel's top partners (Dell, Lenovo, Asus, HP?), followed by more models revealed at CES for release in Q1/Q2 2026.

Yup also only die config will be available For that launch it should be 4+8+4+12 Xe3 Config

 

[Magio]

Yup also only die config will be available For that launch it should be 4+8+4+12 Xe3 Config

Which is good because it's the only die config that interests me in PTL.

Just hoping it performs well on the low end of its TDP limits.

 

[poke01]

Should 4xe3 cores match lunar lakes 8 xe2 cores?

 

[511]

Should 4xe3 cores match lunar lakes 8 xe2 cores?

nope that's too much but maybe they would match 8Xe1 core MTL or slightly less which should be decent imo. btw i am baffled by only 4Xe3 cores should have been 6 Xe3 cores since each render slice is 6 Xe3 cores max.

 

[poke01]

nope that's too much but maybe they would match 8Xe1 core MTL or slightly less which should be decent imo. btw i am baffled by only 4Xe3 cores should have been 6 Xe3 cores since each render slice is 6 Xe3 cores max.

Than wtf is Intel doing?????!

I ain’t paying for the top sku for the 12 GPU cores that’s gotta cost a lot of $$$. Lunar lake comes with 7-8 on the usable SKUs. 226V is 7 core and 256V is 8 core.

I can tell that 12Xe3 SKU will be low volum just like the 288V..

 

[mikk]

Your comparison is not good. 288V is just one SKU whereas 12Xe3 will be included in several SKUs of PTL-H. Furthermore PTL-U is a cheaper solution for the mass whereas LNL-M is a more premium focused product. Intel needs something cheaper than LNL-M. PTL-U is not a LNL-M successor.

 

[Io Magnesso]

As expected, the possibility is low, The XE3 4CORE may have the same performance as the Xe2 8-core…

 

[poke01]

But I thought these were the SKUs?

 

[511]

View attachment 126775

But I thought these were the SKUs?

https://twitter.com/x/status/1924842562902245531

More skus

 

[511]

Your comparison is not good. 288V is just one SKU whereas 12Xe3 will be included in several SKUs of PTL-H. Furthermore PTL-U is a cheaper solution for the mass whereas LNL-M is a more premium focused product. Intel needs something cheaper than LNL-M. PTL-U is not a LNL-M successor.

It's CPU is definitely better than the LNL part the only regressed part is the GPU

 

[poke01]

https://twitter.com/x/status/1924842562902245531

More skus

Oh that’s more like it. The 10xe core sku is good

 

[poke01]

That is probably the Ultra 7 SKU