Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads

[Tigerick]

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.



Comparison of upcoming Intel's U-series CPU: Core Ultra 100U, Lunar Lake and Panther Lake

Model	Code-Name	Date	TDP	Node	Tiles	Main Tile	CPU	LP E-Core	LLC	GPU	Xe-cores
Core Ultra 100U	Meteor Lake	Q4 2023	15 - 57 W	Intel 4 + N5 + N6	4	tCPU	2P + 8E	2	12 MB	Intel Graphics	4
?	Lunar Lake	Q4 2024	17 - 30 W	N3B + N6	2	CPU + GPU & IMC	4P + 4E	0	12 MB	Arc	8
?	Panther Lake	Q1 2026 ?	?	Intel 18A + N3E	3	CPU + MC	4P + 8E	4	?	Arc	12

Comparison of die size of Each Tile of Meteor Lake, Arrow Lake, Lunar Lake and Panther Lake

	Meteor Lake	Arrow Lake (20A)	Arrow Lake (N3B)	Lunar Lake	Panther Lake
Platform	Mobile H/U Only	Desktop Only	Desktop & Mobile H&HX	Mobile U Only	Mobile H
Process Node	Intel 4	Intel 20A	TSMC N3B	TSMC N3B	Intel 18A
Date	Q4 2023	Q1 2025 ?	Desktop-Q4-2024 H&HX-Q1-2025	Q4 2024	Q1 2026 ?
Full Die	6P + 8P	6P + 8E ?	8P + 16E	4P + 4E	4P + 8E
LLC	24 MB	24 MB ?	36 MB ?	12 MB	?
tCPU	66.48
tGPU	44.45
SoC	96.77
IOE	44.45
Total	252.15

Intel Core Ultra 100 - Meteor Lake



As mentioned by Tomshardware, TSMC will manufacture the I/O, SoC, and GPU tiles. That means Intel will manufacture only the CPU and Foveros tiles. (Notably, Intel calls the I/O tile an 'I/O Expander,' hence the IOE moniker.)

 

[poke01]

Please stop with it r23 it's one of the worst benchmarks. At least use 2024 version. In low power SoC MT matters very less. ST is where the focus should be.

No one is rendering blender scenes on a lunar laptop. Browser performance and standby time and video playback are the important metrics here. Throw in some video and photo editing too because these also compete with the base M chips and that's it.

Most of the lunar lake customers won't buy using cinebench scores.

 

[igor_kavinski]

No one is rendering blender scenes on a lunar laptop. Browser performance and standby time and video playback are the important metrics here. Throw in some video and photo editing too because these also compete with the base M chips and that's it.

It should be good for developers too. Almost as good or maybe better than Snapdragon laptops due to better Win32 compatibility. And of course, better Linux support.

 

[511]

ST Responsiveness light MT efficiency and battery life alongside gaming should be where LNL excel at not raw perf we have arrow lake h for that

 

[SiliconFly]

Wut? Every heard of VMs or web?

Nearly every React zoomer runs a Mac. The same applies to your average enterprise Java/Python backend dev.

Wut? Ever heard of native?

Running IDEs like VS on VMs is just a pain due to the emulation overhead. Nothing can beat native.

 

[poke01]

It would be interesting to see how lunar performs in code compilation.

 

[SiliconFly]

It should be good for developers too. Almost as good or maybe better than Snapdragon laptops due to better Win32 compatibility. And of course, better Linux support.

Very true. Running Win32 dev tools native is extremely important. Emulation just doesn't cut it. I used VMs for a while (very painful).

It would be interesting to see how lunar performs in code compilation.

Considering the performance, it should be able to handle small/mid size projects perfectly. If the code base runs into many millions of lines of code, then LL isn't the right choice.

ST Responsiveness light MT efficiency and battery life alongside gaming should be where LNL excel at not raw perf we have arrow lake h for that

Yep 👍

 

[MS_AT]

Very true. Running Win32 dev tools native is extremely important. Emulation just doesn't cut it. I used VMs for a while (very painful).

Could you share the setup? To my knowledge the only problem comes from using VMs to deal with the code that is on the host file system, therefore I am curious what else might go wrong since otherwise dev experience is usually the same if not better in case of poorly configured host.

 

[Nothingness]

Curious question about that. Are these developers fully married to the Mac ecosystem? Whatever binaries they are generating, can only be used on other Macs, if my understanding is correct? Maybe with some slight modification, their open source projects can work on Linux and other BSD derivative distros. And ok, maybe on Windows too with WSL2. But that's it, right? It's a niche of a niche. As far as I understand, they can't be doing anything with their Macs that sells well enough to non-Apple users. That's why I like the idea of Lunar Lake and Snapdragon Elite X. At least it will bring some happiness to regular Windows users who form the majority of the global personal computing userbase.

Among the OS you cite the only one I will never work on is Windows. I never could stand it and I only use it as a gaming OS.

Flash from the past: at the beginning of the century I was working in a startup making a compiler for chip verification; the IT guy only knew about Windows so all our machines were Windows; one day with a collegue we decided we had had enough with that abomination and we just installed Linux. We slowly converted the rest of the team working on the compiler (the GUI team was locked on Windows due to poor initial choice of non portable libraries). The negative impact on the compiler was exactly zero; we just made sure to write portable C++ code.

In my current company, all the engineering work is done on Linux (x86 or Arm). Some people work remotely on cluster/cloud machines (all Linux), so the host OS doesn't matter. I exclusively work locally before dispatching to the cluster, but it's really easy to write portable C/C++ code so developing on macOS (with command line tools, I don't use Xcode) is no issue. I guess I could also work on Windows with WSL2, or keep on doing what I've done for decades, that is working on Linux. But that mac I was lent has been a shock (speed, battery, silence), so I decided I could live with macOS (which I don't like, but it's less hostile to me than Windows).

Most engineers in my company run Linux natively. They start dropping it when they become managers and their life starts revolving around slides. At which point they switch to Windows or Mac (the latter representing the majority).

I don't pretend my experience and preference match the ones from others. I am in a niche, but for sure since my Windows experience at work, one of the most important criteria before changing job was to not have to ever run Windows.

To sum up: I work on portable SW that we have full control over (no non-portable dependencies), so my only need is to have something that looks like *NIX (and macOS with Brew certainly qualifies, as well, I guess, as Windows with WSL2). So my only criteria are laptop quality at large and at the moment Arm-based Apple laptops have my preference.

 

[SiliconFly]

Could you share the setup? To my knowledge the only problem comes from using VMs to deal with the code that is on the host file system, therefore I am curious what else might go wrong since otherwise dev experience is usually the same if not better in case of poorly configured host.

File system performance hence build times vary too much between native and VMs. But, the most important part comes later. Running a test build on a VM isn't the same as running native.

Just out of curiosity, why would any windows developer want to run their ide/tools in a VM rather than native? Doesn't sound right.

 

[yuri69]

Wut? Ever heard of native?

Running IDEs like VS on VMs is just a pain due to the emulation overhead. Nothing can beat native.

VM as VM-based langs. Think of Java VM, Python VM, CLR (.NET VM), etc.

 

[SiliconFly]

Among the OS you cite the only one I will never work on is Windows. I never could stand it and I only use it as a gaming OS.

Flash from the past: at the beginning of the century I was working in a startup making a compiler for chip verification; the IT guy only knew about Windows so all our machines were Windows; one day with a collegue we decided we had had enough with that abomination and we just installed Linux. We slowly converted the rest of the team working on the compiler (the GUI team was locked on Windows due to poor initial choice of non portable libraries). The negative impact on the compiler was exactly zero; we just made sure to write portable C++ code.

In my current company, all the engineering work is done on Linux (x86 or Arm). Some people work remotely on cluster/cloud machines (all Linux), so the host OS doesn't matter. I exclusively work locally before dispatching to the cluster, but it's really easy to write portable C/C++ code so developing on macOS (with command line tools, I don't use Xcode) is no issue. I guess I could also work on Windows with WSL2, or keep on doing what I've done for decades, that is working on Linux. But that mac I was lent has been a shock (speed, battery, silence), so I decided I could live with macOS (which I don't like, but it's less hostile to me than Windows).

Most engineers in my company run Linux natively. They start dropping it when they become managers and their life starts revolving around slides. At which point they switch to Windows or Mac (the latter representing the majority).

I don't pretend my experience and preference match the ones from others. I am in a niche, but for sure since my Windows experience at work, one of the most important criteria before changing job was to not have to ever run Windows.

To sum up: I work on portable SW that we have full control over (no non-portable dependencies), so my only need is to have something that looks like *NIX (and macOS with Brew certainly qualifies, as well, I guess, as Windows with WSL2). So my only criteria are laptop quality at large and at the moment Arm-based Apple laptops have my preference.

I used to work on Linux a lot. But then switched to Windows for some projects. I know VS gets a lot of flak from the community, but once we start using it for a while, it grows on us. And it's obviously the best for Windows.

And most of my code I keep fully portable too (except for parts where it interacts with the system). I too strongly believe, keeping the code fully portable is the right thing to do.

(Sorry abt the rant. I know this is wrong thread for this topic. But just found that post very interesting.)

 

[MS_AT]

File system performance hence build times vary too much between native and VMs. But, the most important part comes later. Running a test build on a VM isn't the same as running native.

Just out of curiosity, why would any windows developer want to run their ide/tools in a VM rather than native? Doesn't sound right.

In one particular case we were using VMs to show IT dept their Windows config (Defender and some other software) was less than optimal, as pure Win install in a VM on the same machine was getting up to 50% better compile times for our projects. Not to mention VM was more consistent with timing of the builds. Then it's also less hassle to install all the tools if you can wrap them in Win container [Win on Win] but this is not always feasible.

 

[SiliconFly]

In one particular case we were using VMs to show IT dept their Windows config (Defender and some other software) was less than optimal, as pure Win install in a VM on the same machine was getting up to 50% better compile times for our projects. Not to mention VM was more consistent with timing of the builds. Then it's also less hassle to install all the tools if you can wrap them in Win container [Win on Win] but this is not always feasible.

I used to work on a system software project where using VMs wasn't an option for debugging and testing. And we used to mess up the systems constantly during development/testing. Happens all the time.

But it was very easily addressed with imaging. Even in case of a full corruption it can bring the system back up to spec in less than 10 mins, byte-for-byte. Configuring/deploying images is as easy as maintaining/deploying VMs once the admins get the hang of it. I'm in so much love with imaging, I even use it for my personal laptops even today.

 

[Tup3x]

It should be good for developers too. Almost as good or maybe better than Snapdragon laptops due to better Win32 compatibility. And of course, better Linux support.

I don't know about that, especially if you happen to be a web dev (where pretty much everything is available for ARM already). X Elite does have advantage in multitasking and might be faster when running WSL2 + VS Code + gazillion browser windows/tabs. At times my current laptop (T14 Gen 1, Ryzen 7 PRO 4750U) is dog slow.

I don't think Lunar Lake would quite cut it. Without HT I'd want at least 12 cores, 32GB and higher TDP than 17W.

File system performance hence build times vary too much between native and VMs. But, the most important part comes later. Running a test build on a VM isn't the same as running native.

Just out of curiosity, why would any windows developer want to run their ide/tools in a VM rather than native? Doesn't sound right.

Compatibility reasons. Since most people in my team use Macs, things will just work if I run stuff in WSL2. Otherwise I'd have to make Windows specific build scripts for projects.

 

[SiliconFly]

It appears that some of us may have derailed the topic from Lunar Lake performance to dev tools. Think it's better to switch back to the original topic asap before people start noticing.

 

[igor_kavinski]

I don't think Lunar Lake would quite cut it. Without HT I'd want at least 12 cores, 32GB and higher TDP than 17W.

288V SKU is 30W and 32GB RAM.

Another option I've posted in the GB6 thread.

 

[DavidC1]

At times my current laptop (T14 Gen 1, Ryzen 7 PRO 4750U) is dog slow.

Lunarlake has a huge single thread and IO(memory/drive) system advantage. It helps worst case scenarios more than on average.

Read many instances of people saying how much Zen 3 was over Zen 1/2 or Skylake over Haswell or Alderlake vs Coffeelake. Usually min fps in gaming being a huge gain.

Computing advances always favor the lower power, lower cost, smaller devices more than the opposite.

 

[majord]

No it doesn't.

From Geekerwan:

View attachment 106157

I don't see the 6800/6850U in this graph?

 

[ikjadoon]

It would be interesting to see how lunar performs in code compilation.

GB6 has a Clang compilation sub-score.

Clang
The Clang workload uses the Clang compiler to compile the Lua interpreter, a popular opensource language interpreter. It models the use case of developers building their code and the just-in-time compiling that general users can encounter on their systems (such as JIT compilation for scripting Java and compilation for shading languages in GPU drivers). This workload uses the musl libc as the C standard library for the compiled files.The Clang workload compiles 8 files in single-core mode and 96 files in multi-core mode.

The 268V scores 14514 nT points (71.5K lines / second) in the Clang subscore. I don't know which HX 370 laptops use a lower power limit, but the lower-wattage RPL Refresh SKUs (e.g., 2P+8E) like the 150U Clang subscore is 11661 nT points (57.4 Klines/sec).

The HX 370's Clang subscore whacks them both with 25474 nT (125.5 Klines/sec) at an unknown wattage.

288V SKU is 30W and 32GB RAM.

Another option I've posted in the GB6 thread.

I'm hopeful LNL mini-PCs allow a 'TDP Up' config of PL1 = PL2 = 30W on any SKU. That 288V (a Core Ultra 9) may be quite pricey and not offered by most OEMs.

ASUS Lunar Canyon reference NUC looks like it can cool 30W sustained on any SKU.

 

[DavidC1]

I don't see the 6800/6850U in this graph?

Why would Hawx and Strix underperform 6800/6850U?

That's the point. Not even the Phoenix 8945H does 10.2K at 23W. The implication would be then that AMD regressed noticeably since Remembrandt in early 2022.

 

[poke01]

GB6 has a Clang compilation sub-score.



The 268V scores 14514 nT points (71.5K lines / second) in the Clang subscore. I don't know which HX 370 laptops use a lower power limit, but the lower-wattage RPL Refresh SKUs (e.g., 2P+8E) like the 150U Clang subscore is 11661 nT points (57.4 Klines/sec).

The HX 370's Clang subscore whacks them both with 25474 nT (125.5 Klines/sec) at an unknown wattage.



I'm hopeful LNL mini-PCs allow a 'TDP Up' config of PL1 = PL2 = 30W on any SKU. That 288V (a Core Ultra 9) may be quite pricey and not offered by most OEMs.

ASUS Lunar Canyon reference NUC looks like it can cool 30W sustained on any SKU.

So not that impressive relative to the current competition. Disabling Hyper threading is hurting them here. I think Intel should have left that as an option. Lion cove isn’t the latest the ARM core where SMT isn’t needed. It still needs HT but I guess they removed it to show ST improvements.


Thanks for that.

 

[Hulk]

Cinebench has nothing to do with the software I actually run on my computers.

But it is a good benchmark for me because I know by heart how most chips perform with CB, and that in turn gives me a very accurate idea of how they perform in application that are important to me.

For example, my currrent laptop does about 3200 in CB R23 MT. A laptop scoring 10,000 in that bench will be about 3x faster than my current laptop.

CB is meaningless and full of meaning at the same time

 

[DrMrLordX]

Why would Hawx and Strix underperform 6800/6850U?

That's the point. Not even the Phoenix 8945H does 10.2K at 23W. The implication would be then that AMD regressed noticeably since Remembrandt in early 2022.

Maybe look for benches of the 6850u instead since that data is widely available:

https://www.notebookcheck.net/AMD-Ryzen-7-PRO-6850U-Processor-Benchmarks-and-Specs.615519.0.html

Sorry to be "that guy" but this is an Intel thread, and not a thread for old AMD mobile SoCs.

Cinebench has nothing to do with the software I actually run on my computers.

But it is a good benchmark for me because I know by heart how most chips perform with CB, and that in turn gives me a very accurate idea of how they perform in application that are important to me.

For example, my currrent laptop does about 3200 in CB R23 MT. A laptop scoring 10,000 in that bench will be about 3x faster than my current laptop.

CB is meaningless and full of meaning at the same time

It used to be a good fp benchmark. Now I'm not so sure, since fp has become so complicated with all the ISA extensions out there you can use, and some uarches do better with old stuff like SSE4.x than others, etc. Benchmarking has changed over the years.

 

[poke01]

I don’t get why some people show old AMD SKUs. There is nothing like Lunar Lake on the AMD side. It’s honestly the most advanced laptop x86 design in a long time and it will only be beaten by Strix Halo.

Yes, your old AMD U SKU will score much higher in cinebench but I doubt the power consumption is the same as well the responsiveness.

I bet that that Lunar will also beat the X Elite in battery life.

 

[majord]

Why would Hawx and Strix underperform 6800/6850U?

That's the point. Not even the Phoenix 8945H does 10.2K at 23W. The implication would be then that AMD regressed noticeably since Remembrandt in early 2022.

Probably because the H SKU isn't optimised for thosecTDPs? I dunno, but that's not representative of U series such as 7840U

e.g:

Acer Swift Edge 16 OLED review (SFE16-43 model, AMD Ryzen 7 7840U)

This is my detailed review of the Acer Swift Edge 16 SFE16-43 generation, a lightweight and portable 16-inch laptop built on a modern AMD Ryzen U hardware

www.ultrabookreview.com

10min loop settles on around 11K CB23 @ 17W package in their CB23 loop screenshot

Since many not convinced, here.. tested right now, not even clean run on a 2 yr old O/S

Hwinfo cleared at start of run for ave package power. Screenshot taken instant bench score appears:

24w avg: 9920cb
19w avg (balanced) 9089cb
12w avg: (power saver) 7088cb








and incase there's any doubt about package pwr readings..

5min loop 18.4w package, 22.8w battery draw w/ brightness low, wifi on