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	8 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)	Arrow Lake Refresh (N3B)	Lunar Lake	Panther Lake
Platform	Mobile H/U Only	Desktop Only	Desktop & Mobile H&HX	Desktop Only	Mobile U Only	Mobile H
Process Node	Intel 4	Intel 20A	TSMC N3B	TSMC N3B	TSMC N3B	Intel 18A
Date	Q4 2023	Q1 2025 ?	Desktop-Q4-2024 H&HX-Q1-2025	Q4 2025 ?	Q4 2024	Q1 2026 ?
Full Die	6P + 8P	6P + 8E ?	8P + 16E	8P + 32E	4P + 4E	4P + 8E
LLC	24 MB	24 MB ?	36 MB ?	?	8 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.)

 

[igor_kavinski]

Wasn't it a given that actual availability of Meteor Lake in Q4 2023 would be nearly non-existent and very expensive?

True. But I get why the OEMs might be furious. Intel announced MTL officially so a lot of people may have put their laptop buying plans on hold to get the latest Intel thing. So not only do the OEMs not have MTL laptops to sell but what they do have in the channel may also see decreased sales. Intel double whammied them.

 

[dullard]

Intel announced MTL officially so a lot of people may have put their laptop buying plans on hold to get the latest Intel thing.

What, I thought Intel has been silent?

 

[poke01]

Geekbench Search - Geekbench

browser.geekbench.com

So what you guys make of these results?

My take, ST behind AMD and Apple. MT on par with M3 base. This is Intel's midrange chip for ultrabooks in the first half of 2024.

 

[igor_kavinski]

Geekbench Search - Geekbench

browser.geekbench.com

Dynabook Inc. PORTEGE X40L-M vs Intel Corporation Meteor Lake Client Platform - Geekbench

browser.geekbench.com

Wild perf variation between OEM and Intel reference mobo/laptop. Could be due to having set lower power limits in the BIOS or that OEM has no idea how to cool the chip properly.

 

[jpiniero]

Dynabook Inc. PORTEGE X40L-K vs Dynabook Inc. PORTEGE X40L-M - Geekbench

browser.geekbench.com

Not really faster compared to a Dynabook with the 1370P.

 

[poke01]

Dynabook Inc. PORTEGE X40L-M vs Intel Corporation Meteor Lake Client Platform - Geekbench

browser.geekbench.com

Wild perf variation between OEM and Intel reference mobo/laptop. Could be due to having set lower power limits in the BIOS or that OEM has no idea how to cool the chip properly.

It's not a cooling problem. I think these Meteor Lake ultrabooks aren't much faster than Raptor mobile.

They might be better at a higher wattage.

 

[Hulk]

Comparing mobile parts even after official release is hard enough because every system throttles to some extent. Trying to compare them before release introduces even greater variability since we don't know if these are retail parts QS, or whatever, we don't know clocks, memory subsystems, etc... I suggest we wait until release before we start arguing about what a giant fail or success this is for Intel.

 

[igor_kavinski]

Dynabook Inc. PORTEGE X40L-K vs Dynabook Inc. PORTEGE X40L-M - Geekbench

browser.geekbench.com

Not really faster compared to a Dynabook with the 1370P.

Damn. MTL really suffers in Ray Tracing, HDR, Clang and Asset compression. Could be an artifact of pretty bad memory latency or lower RAM bandwidth.

BIOS at ver 0.20 though. Their intern is doing their best to understand the chip!

 

[poke01]

Comparing mobile parts even after official release is hard enough because every system throttles to some extent. Trying to compare them before release introduces even greater variability since we don't know if these are retail parts QS, or whatever, we don't know clocks, memory subsystems, etc... I suggest we wait until release before we start arguing about what a giant fail or success this is for Intel.

Yep, will reserve my opinions till release.

 

[SiliconFly]

A

It's been like that pretty much through the entire Zen era. AMD ships some mobile SoCs but you never really know where they're going. Availability at retail is bad for months on end.

Actually, you can find excellent AMD laptops in Alienware, Razer Blade, Asus ROG, Acer Predator, Nitro, HP Omen, etc.

 

[Thibsie]

A

Actually, you can find excellent AMD laptops in Alienware, Razer Blade, Asus ROG, Acer Predator, Nitro, HP Omen, etc.

Nah, those are gamer bricks using heavier PSU than the laptop itself (yeah, I pushed it but you get the idea).
Thin & Light yet performing (relative to their thin & light constitution of course) is a lot harder.

 

[tamz_msc]

Geekbench 6 is calibrated to give 2500 score on a desktop 12700 in the ST tests.

Since it is performing essentially the same as a desktop 12700, considering the variability of Geekbench runs on a laptop, I would say it's performing as expected.

The MT score is really impressive, considering 8+4 for the 12700 vs 6+8 for the Meteor Lake chip, with the obvious caveat that it's not known what the LP Island E-cores do when the CPU is run at full tilt. Lower boost frequency of the E-cores in general may also be a factor.

 

[igor_kavinski]

with the obvious caveat that it's not known what the LP Island E-cores do when the CPU is run at full tilt.

I read that they generally do not take part in MT workloads unless you force them to participate through process affinity. Even then, their contribution to the MT performance is nothing to write home about.

 

[igor_kavinski]

Dynabook Inc. PORTEGE X40L-K vs Intel Corporation Meteor Lake Client Platform - Geekbench

browser.geekbench.com

If the reference laptop is any indication of final performance, there is "some" hope for MTL to not be a failure. Notice the MTL chip being on Stepping 4. Got delayed by silicon bug fixes, I guess?

Not that bad compared to a 13700H but still getting beat pretty badly in some tests: https://browser.geekbench.com/v6/cpu/compare/3435865?baseline=2772108

 

[mikk]

ST is worse on MTL. Because of the same IPC and roughly ~300 Mhz less peak frequency it is not a surprise anymore. Also the tile architecture seems to introduce some extra latency over Raptor Lake which contributes to it. ST isn't the most important if it's just a 5% slowdown. Efficiency/Battery life matters much more.

 

[Thunder 57]

Dynabook Inc. PORTEGE X40L-K vs Intel Corporation Meteor Lake Client Platform - Geekbench

browser.geekbench.com

If the reference laptop is any indication of final performance, there is "some" hope for MTL to not be a failure. Notice the MTL chip being on Stepping 4. Got delayed by silicon bug fixes, I guess?

Not that bad compared to a 13700H but still getting beat pretty badly in some tests: https://browser.geekbench.com/v6/cpu/compare/3435865?baseline=2772108

The fact that is loses the single threaded tests but does pretty well in MT makes me think it is more power efficient and able to maintain higher boost than RPL.

 

[Thibsie]

Dynabook Inc. PORTEGE X40L-K vs Intel Corporation Meteor Lake Client Platform - Geekbench

browser.geekbench.com

If the reference laptop is any indication of final performance, there is "some" hope for MTL to not be a failure. Notice the MTL chip being on Stepping 4. Got delayed by silicon bug fixes, I guess?

Not that bad compared to a 13700H but still getting beat pretty badly in some tests: https://browser.geekbench.com/v6/cpu/compare/3435865?baseline=2772108

IMO MTL is more about going to a modular approach with more or less the same stuff than anything else.
I'm not pointing the finger at Intel, quite the opposite: it is wise not to do to much at the same time.

If this is a success on a manufacturing PoV (including yield etc) : good. Intel will move to the next goal.

It is clear though than from a pure performance PoV, MTL isn't earth shattering to say the least but IMO this wasn't the point at all. And it is OK to me.

 

[tamz_msc]

Also regarding power consumption - the 12700 has a PL2 of 180 W. Considering the 28-35 W TDP of these Meteor Lake CPUs for which I really doubt that PL2 would be set much higher than 65 W (which is a reasonable assumption given that RPL-P chips have their PL2 set at 64 W as per spec), we are potentially seeing ~1/3 power consumption at iso-frequency and similar performance levels.

 

[Tuna-Fish]

3, 3 wide decoders sounds pretty weird tbh. And one of the decoders is disabled for a total of 8 wide?
Maybe they want SKMT to target extremely low power so they could power gate to just 2 or 3 wide decoders if they want to, idk

I would assume the decoders are 9 wide total, but something right after them is 8 wide. 3x3 decode doesn't get you 9 ops every clock, there will be mispredictions etc. If there's room to buffer a bit between them, if you have for example 8 wide rename, going 9 wide in the decode still probably win you a little, even if it's impossible to maintain 9 ops per cycle every cycle.

 

[eek2121]

The fact that is loses the single threaded tests but does pretty well in MT makes me think it is more power efficient and able to maintain higher boost than RPL.

Of course it is more efficient. All these people are dooming it without understanding that the chip sips power.

Also, this isn’t even the fastest SKU. Get back to me when the 185 sku has a good GB run.

 

[jpiniero]

Of course it is more efficient. All these people are dooming it without understanding that the chip sips power.

Same thing happened with Ice Lake. I'm guessing they loosened the bins as far as possible to get enough chips. So it may not be more efficient than Raptor Lake.

 

[tamz_msc]

Same thing happened with Ice Lake. I'm guessing they loosened the bins as far as possible to get enough chips. So it may not be more efficient than Raptor Lake.

Intel 10 was in a much worse shape than Intel 4. Ice Lake didn't even have 4 GHz turbo if I'm not mistaken.

 

[SpudLobby]

Do we have any more recent indication as to what kind of volume MTL might be shipping? Meme volume sure but anything quantitative or guesses?

Also *when*, particularly for < 35W. I’m guessing we’ll see Q1-Q3 ‘24 for serious availability.

 

[SpudLobby]

IMO MTL is more about going to a modular approach with more or less the same stuff than anything else.
I'm not pointing the finger at Intel, quite the opposite: it is wise not to do to much at the same time.

If this is a success on a manufacturing PoV (including yield etc) : good. Intel will move to the next goal.

It is clear though than from a pure performance PoV, MTL isn't earth shattering to say the least but IMO this wasn't the point at all. And it is OK to me.

Agree RE. Performance. Efficiency under load and idle efficiency are just far more important right now

 

[SpudLobby]

A

Actually, you can find excellent AMD laptops in Alienware, Razer Blade, Asus ROG, Acer Predator, Nitro, HP Omen, etc.

Those are all gaming or beefy laptops. I hope this is facetious about how piss poor AMD is on mobile laptops 15-35W that the average person might want to use lol. It’s still atrocious honestly especially in the US.