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.

 

[adroc_thurston]

BTW, I meant Dell (rather than HP) not offering Strix Halo.

Dell barely started offering AMD in normal commercial.

Intel did other things, such as added LP cores, added NPU (in order to stay in corporate PCs), improved the iGPU.

that's nice. where's the CPU perf? this is 2026, not 2023. QCOM is breaching 4k GB6 1t.

Which makes it even more important that the design printed on the silicon is the most competitive, not stale.

no, it means you gotta run tf away from that market until you're out of breath.

 

[regen1]

https://twitter.com/x/status/2015882390569169187

 

[QuickyDuck]

We finally get big LNL and it only takes 1 year!
What can you ask more?

 

[511]

you mean 10?

View attachment 137390
test conditions:

• Test method: background apps logged into QQ, Chrome with Office tabs open
• Browser and QQ kept running
• Same Wi-Fi
• Brightness: 200 nits
• Power mode: “Balanced / Default”

Lunar is better, interesting

What about the panel is it OLED?

 

[poke01]

What about the panel is it OLED?

both lunar and panther lenovos are oled

 

[regen1]

Valid criticism of Microsoft/Windows environment ~3:18 onwards.

 

[mikegg]

I halved my CPU's clockspeed in BIOS, and ran the test again. The LCP took 780ms, which is only took 7% longer than stock. So it looks like chrome devtools CPU throttling is inaccurate, just not in the direction you expected.

For GPU rendering, input, OS scheduling, etc. I don't see your point. We are discussing Intel Laptop CPUs scoring lower in javascript benchmarks. They might also throttle GPU or have bad input latency, but I've seen no reputable claims for that.

That’s because clock speed and performance isn’t linear. Chrome dev tools literally divide the JS performance by 4.

For GPU rendering, it’s relevant because we are talking about battery life while browsing here. Intel is going to throttle the entire chip, CPU and GPU.

 

[511]

it's better for intel to keep working on a in house Linux Distro for their hw and support it and focus on it rather than rely on microsoft same for AMD microslop has been making windows bad

 

[511]

That’s because clock speed and performance isn’t linear. Chrome dev tools literally divide the JS performance by 4.

For GPU rendering, it’s relevant because we are talking about battery life while browsing here. Intel is going to throttle the entire chip, CPU and GPU.

throttling is power profile dependent you don't get throttle on performance mode you get it on balanced so there is that

 

[mikegg]

throttling is power profile dependent you don't get throttle on performance mode you get it on balanced so there is that

Trace the replies back. No Youtuber ever tests battery life on performance mode. It's always the default mode.

PS. Windows does throttle on performance mode as well.

 

[regen1]

Valid criticism of Microsoft/Windows environment ~3:18 onwards.

https://twitter.com/x/status/2015948530578587699

 

[511]

Trace the replies back. No Youtuber ever tests battery life on performance mode. It's always the default mode.

PS. Windows does throttle on performance mode as well.

it doesn't?

 

[coercitiv]

PS. Windows does throttle on performance mode as well.

Let's do the data thing:

https://www.laptopmag.com/ai/copilot-pcs/snapdragon-x-elite-chips-battery-power-amd-intel

Header Cell - Column 0	Qualcomm Snapdragon X Elite: Plugged in (Best Performance)	Qualcomm Snapdragon X Elite: On battery (Best Performance)	Qualcomm Snapdragon X Elite: On battery (Balanced)	Intel Core Ultra 200V: Plugged in (Best Performance)	Intel Core Ultra 200V: On battery (Best Performance)	Intel Core Ultra 200V: On battery (Balanced)	AMD Ryzen AI Max+ Pro 395: Plugged in (Best Performance)	AMD Ryzen AI Max+ Pro 395: On battery (Best Performance)	AMD Ryzen AI Max+ Pro 395: On battery (Balanced)
Geekbench 6: Single core (Higher is better)	2,379	2,283	2,309	2,721	2,699	1,599	2,837	2,647	2,224
Geekbench 6: Multicore (Higher is better)	12,340	9,207	7,589	11,035	10,988	9,058	17,721	14,244	13,416
Cinebench R23: Single core (Higher is better)	1,083	1,073	1,104	1,829	1,891	1,033	1,930	1,940	1,050
Cinebench R23: Multicore (Higher is better)	10,130	11,102	9,941	8,431	7,952	7,875	29,469	24,208	19,733
Handbrake video encoding (Lower is better)	06:50	11:30	08:09	07:26	08:50	08:44	02:36	03:27	03:37
3DMark Fire Strike (Higher is better)	5,800	5,795	4,965	8,462	7,412	7,787	23,459	16,151	15,593
3DMark Time Spy (Higher is better)	1,873	1,891	1,803	3,896	3,557	3,735	10,114	7,362	6,689

Let's take a look at GB6 Multicore, which is quite representative for a mixed workload running on a laptop (not ST, not massive MT either).


Performance retention on battery /w Balanced profile in GB6 MT:
Snapdragon X Elite: 61.5%
Intel Lunar Lake: 82%

Performance retention on battery /w Best Performance profile in GB6 MT:
Snapdragon X Elite: 74.6%
Intel Lunar Lake: 99.6%

Spoiler: The max performance you deserve

Intel has tried quite a few times in the past to dunk on AMD for the exact thing we're discussing here, they claimed their chips were superior because they offered more consistent performance on battery vs AMD. People didn't really care, users are willing to sacrifice some performance for stamina on battery as long as performance is still adequate. (ideally we want user choice, I like how MacOS does this with very good performance retention on default and additional profiles for better battery life and/or noise/thermals)

The part that really sucks here is Window's implementation of power profiles and power settings that need to be changed in multiple places. One good thing they accomplished somewhat recently was to introduce both "Plugged-in" and "On battery" dropdowns for power modes, in the past there was only a "unified" selector that changed value depending on whether the unit was on battery or plugged in, which meant inexperienced reviewers did not even know what profile they were testing with.

Look at this, and prepare to facepalm:

 

[regen1]

酷睿Ultra 5 338H首发测试报告：核显小王降临，压力给到友商_哔哩哔哩_bilibili

时间有限简单测试了一下338H，10Xe的表现可以说是相当惊艳，相比于满血的12Xe性能损失相当小，相比于285H的140T能够实现超过50%的性能提升。, 视频播放量 3772、弹幕量 7、点赞数 175、投硬币枚数 52、收藏人数 47、转发人数 38, 视频作者 金猪升级包, 作者简介 既不独立，也不客观，更不公正，相关视频：独显下岗了？核显200多帧玩3A：还有超长续航？Panther Lake版 惠普星Book Pro 16 2026——Ultra X7 358H全球首测！，【首发评测】Intel 酷睿Ultra X9 388H，联想小新Pro 16 GT...

www.bilibili.com

Some U5 338H(B370/10Xe3) benchmarks from "Goldenpig Upgrade"

Gaming and Synthetics:










Cinebench:




some "AI"

 

[Covfefe]

That’s because clock speed and performance isn’t linear. Chrome dev tools literally divide the JS performance by 4.

I ran speedometer stock and after throttling in bios. I scored 39.6 and 21.2. conveniently that lines up exactly with the 54% that panther lake scores on battery.

For GPU rendering, it’s relevant because we are talking about battery life while browsing here. Intel is going to throttle the entire chip, CPU and GPU.

The fact that they're throttling CPU in certain applications doesn't mean they're throttling GPU. I'm sure they're power limiting it, but so is AMD and Apple and Qualcomm. There's no indication that Intel is throttling the GPU enough to make the experience choppy. If they were, reviewers would have mentioned it.

 

[DavidC1]

Ultra 5 338H is the star of the show. 85% of graphics performance. Still 50% faster than predecessor.

Trace the replies back. No Youtuber ever tests battery life on performance mode. It's always the default mode.

PS. Windows does throttle on performance mode as well.

Balanced isn't necessarily slower than performance. The Dell XPS 12 I had ran Starcraft 2 better on battery Balanced than on Performance mode. Noticeably actually. In the 25% range.

 

[DavidC1]

you mean 10?

I mean 8. There's a video where it compares all those generations. 8.1, 10 and 11. It's progressively more bloated, more memory use and such.

Also, laptop battery life will vary depending on vendor, implementation and tested settings too. I doubt the differences will be noticeable overall. I would bet the dual screen implementation means it has some overhead even if one screen is off. Laptops are finicky like that.

 

[regen1]

Balanced isn't necessarily slower than performance. The Dell XPS 12 I had ran Starcraft 2 better on battery Balanced than on Performance mode. Noticeably actually. In the 25% range.

For gaming something similar was observed here, around 13 minute :

For this particular test it was more performant, quieter in "Balanced" mode.

 

[DavidC1]

For this particular test it was more performant, quieter in "Balanced" mode.

I assume it's because how Turbo 2.0(Sandy Bridge) works, which uses unspent thermal headroom of a cooling system. Performance runs full throttle all the time, thus the heatsink is in average warmer, while Balanced mode allows the SoC to cool down having a reverse effect of being faster.

 

[coercitiv]

No Youtuber ever tests battery life on performance mode. It's always the default mode.

The Dell XPS 12 I had ran Starcraft 2 better on battery Balanced than on Performance mode. Noticeably actually. In the 25% range.

Dell XPS 13 Elite 1st gen versus LNL, custom battery test with developers workload, max screen brightness and AFAIK Best Performance power mode (as claimed previously during the video):

The Lunar Lake machine outlasted the X Elite model. Now, we could argue that we need to establish a proper baseline for such a test, such as exact same display panel and battery, but like I already hinted in my previous post, even the power modes are not the same on many Windows devices.

 

[mikegg]

Dell XPS 13 Elite 1st gen versus LNL, custom battery test with developers workload, max screen brightness and AFAIK Best Performance power mode (as claimed previously during the video):

The Lunar Lake machine outlasted the X Elite model. Now, we could argue that we need to establish a proper baseline for such a test, such as exact same display panel and battery, but like I already hinted in my previous post, even the power modes are not the same on many Windows devices.

These tests are meaningless without the amount of work done.

The Qualcomm version could have done 50% more work before it ran out of battery first. X Elite doesn't throttle (or not nearly as much as LNL) and it has much higher MT performance. So it could very well have done much more work than LNL in the same time.

 

[DavidC1]

These tests are meaningless without the amount of work done.

The Qualcomm version could have done 50% more work before it ran out of battery first. X Elite doesn't throttle (or not nearly as much as LNL) and it has much higher MT performance. So it could very well have done much more work than LNL in the same time.

@coercitiv Just showed you that Snapdragon throttles more than Lunarlake in certain implementations.

 

[mvprod123]

 

[DavidC1]

I would think at this point, a separate GPU oriented SKU for desktop with 12 Xe cores might have been interesting. Say a Core Ultra 5 245 had that for $430. That's $130 over the normal chip. On desktops DDR5 goes quite high clocks.

View attachment 137390
Lunar is better, interesting

I looked closer. The 388H is getting 7.7 with 120Hz display. Lunarlake is getting 7.9, so that's a minor difference, not even 5%.

Intel Panther Lake Arc B390 performance and efficiency analysis: Intel's new iGPU trades blows with the Nvidia GeForce RTX 4050

Notebookcheck analyses the new Intel Arc B390 integrated graphics in the Panther Lake Core Ultra X9 388H powering the Asus Zenbook Duo UX8407

www.notebookcheck.net

The efficiency comparison becomes interesting with lower TDP values for the Arc B390. The efficiency is almost at the level of the Apple M5 GPU even at 35 W, and this gets even better at 28 W and 20 W.

Same process, same efficiency. GPUs are embarassingly parallel so much easier to catch up, hence why Apple doesn't get decisive advantage in GPU vs competitors.

 

[coercitiv]

X Elite doesn't throttle (or not nearly as much as LNL) and it has much higher MT performance.

I have already shown X Elite throttles in some laptop implementations. When using the Best Performance power mode, the Yoga Slim 7x retains only ~75% performance when switching to battery. We can also see the Balanced mode is throttling the X Elite more than Lunar Lake.

A big part of this throttling is the OS and OEM custom implementation, so this will vary a lot from one device to another.