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.)

 

[Hitman928]

Is ARL's tGPU based on an updated version MTL's Alchemist tGPU? Or is it based on Battlemage?

Recently, I read an article that said ARL's tGPU is Battlemage. I remember reading older articles that said it's based on Alchemist++. Any idea which one it's gonna be?

Should be Alchemist based.

 

[ikjadoon]

PC OEMs appear to have quite buggy prototypes. The MSI Meteor Lake prototype provided by Intel for media benchmarking was on the struggle bus today:

During testing, the Meteor Lake laptops BSOD'd, froze up on 1T tests, hit 116C as MSI/Intel launched with an asinine 64W PL1 & 115W PL2 in a 16" thin chassis. CPU power hit 80W, immediately throttled.

Why would Intel give these devices to media? Looks like a rushed release.

 

[cebri1]

Why would you have an event to overclock laptop CPUs meant to provide a jump in efficiency. Intel is a master at shooting at their own foot. Pretty horrible showing.

 

[Hitman928]

Why would you have an event to overclock laptop CPUs meant to provide a jump in efficiency. Intel is a master at shooting at their own foot. Pretty horrible showing.

They're not overclocking, that's how the OEM has it configured out of the box. TDP ratings on mobile CPUs are really just a vague suggestion today.

 

[ikjadoon]

Why would you have an event to overclock laptop CPUs meant to provide a jump in efficiency. Intel is a master at shooting at their own foot. Pretty horrible showing.

That lower-thirds stating "Laptop OC Event" was weird. H-series CPUs are traditionally locked and it's not visibly overclocked in Hwinfo64: 5 GHz is the stock boost clock of the 165H.

I think it's just genuinely unstable, super-hot, and throttling at stock clocks—at least in that chassis / design.

//

Maybe Intel is "redefining" overclocking, the event name by KitGuru is off, or maybe this event also featured the 14th Gen Raptor Lake HX mobile CPUs, which have unlocked multipliers and were rumored to launch at CES 2024.

The Geekbench entry gives us some insight into the capabilities of the forthcoming i7-14700HX CPU, with a maximum clock speed of 5,287MHz. However, additional data from this test indicates that the clock frequency reached even higher levels, peaking at 5462MHz, putting it into the 5.4–5.5GHz range.

 

[Kepler_L2]

Is ARL's tGPU based on an updated version MTL's Alchemist tGPU? Or is it based on Battlemage?

Recently, I read an article that said ARL's tGPU is Battlemage. I remember reading older articles that said it's based on Alchemist++. Any idea which one it's gonna be?

Alchemist LPG+

 

[trivik12]

Intel has released 3 new cpus Core 7,Core 5. Core 4 (no ultra) today. Initially I thought they are MTL because of not calling them 14th gen. But now I think they are Raptor Lake as there are no LP E-cores in it.

Intel Core Processors Series 1 launched - new 15W U chips for low-power systems

Launching in laptops during Q1 2024

www.tomshardware.com

This is super annoying. New naming with old cores.

 

[jpiniero]

Intel has released 3 new cpus Core 7,Core 5. Core 4 (no ultra) today. Initially I thought they are MTL because of not calling them 14th gen. But now I think they are Raptor Lake as there are no LP E-cores in it.

Raptor Lake is gonna be the majority of laptop volume for Intel in 2024 and probably 2025. May as well rebrand.

 

[Dayman1225]

Intel has released 3 new cpus Core 7,Core 5. Core 4 (no ultra) today. Initially I thought they are MTL because of not calling them 14th gen. But now I think they are Raptor Lake as there are no LP E-cores in it.

Intel Core Processors Series 1 launched - new 15W U chips for low-power systems

Launching in laptops during Q1 2024

www.tomshardware.com

This is super annoying. New naming with old cores.

That’s why Ultra exists, to be able to tell the difference between newest architecture vs older

 

[Henry swagger]

https://wccftech.com/intel-arrow-la...l-next-gen-lunar-lake-significant-ipc-3x-npu/. Intel teasing arrow lake amd lunar lake.. arrow will come with accelerators like xeon cpu's

 

[DAPUNISHER]

I'm excited for a Lunar Lake desktop APU with Gandalf/Battlemage. Also stoked Intel is coming out swinging in the handheld market. And getting back to form in the mini/NUC arena. If performance per watt is on point? AMD better bite down on their mouthpiece, it'll be barn burner for the next few rounds.

 

[Geddagod]

That s made up numbers, he use the package power, so the uncore power is included in the alleged core power comsumption, that s why the curves are that steep at the beginning, by increasing artificially the idle power he shift the curves where it stand his doctored numbers.

Yet HVinfo has power per core data, one has to wonder why he didnt isolate a single core power using this tool, likely that it would had contradicted his purposely biaised numbers.

Ur only half wrong this time, bravo.
Yes using pckg power isn't as good as using just core power, but he never pretended to do otherwise. He actually also does do core power tests as well, just not for MTL yet.
"made up numbers" "doctored" "artificially increasing idle power" my butt lmao, the cope here is sad.
He also shows RWC as having a worse curve than Zen 4, and the abysmal SOC cores curve. He even didn't even include the 7840hs curve which is much worse than the 7840u curve. But because one core curve (crestmont) is better than AMD at an SOC power <5 wats, he has purposefully biased the numbers lmfao. Ok buddy.

 

[Geddagod]

PC OEMs appear to have quite buggy prototypes. The MSI Meteor Lake prototype provided by Intel for media benchmarking was on the struggle bus today:

During testing, the Meteor Lake laptops BSOD'd, froze up on 1T tests, hit 116C as MSI/Intel launched with an asinine 64W PL1 & 115W PL2 in a 16" thin chassis. CPU power hit 80W, immediately throttled.

Why would Intel give these devices to media? Looks like a rushed release.

Yikes. Hopefully other OEMs are slightly better. I had an MSI laptop before with a 9750H iirc, and it throttled like sh*t too, and the hinge broke in a couple years lol.
The MTL launch so far has been a major L overall tho. Quite a shame.

 

[Geddagod]

LNL gets shown off. Compute die, SOC die, and rumored dummy structural die.
Area efficiency doesn't look too great TBH. The compute die alone on N3 looks to be the same size as the M3 die.

 

[Thunder 57]

Yikes. Hopefully other OEMs are slightly better. I had an MSI laptop before with a 9750H iirc, and it throttled like sh*t too, and the hinge broke in a couple years lol.
The MTL launch so far has been a major L overall tho. Quite a shame.

No kidding. When will Intel get their ...crap? together? I have zero faith in ARL at this point.

 

[coercitiv]

The compute die alone on N3 looks to be the same size as the M3 die.

Nonsense, Intel just needs to pick employees with bigger hands to show the chips.

I don't know the exact size of M3, but I assume we're talking about ~150mm2 or less. IRRC this seems to be a size both Intel and AMD tried to hit in the past with their mobile dies, in the sense that they usually made sacrifices to stay around this mark (core count, cache etc). Therefore, for good or worse, this was likely the intended area target.

BTW, is there clear info on LNL cache structure? I remember seeing the diagram with 8MB "System Cache" but no L3 mentioned, and the Sisoft Sandra submission wasn't much help either.

 

[DavidC1]

@coercitiv L3 was mentioned. "LLC", but only for the P. The E cores are basically LPe cores and only come with their native L2 cache.

Lunarlake actually shows some promise unlike Meteorlake's asterick here and asterick there for getting good battery life. The system level cache is for reducing power use, which is akin to what Apple chips do too.

I don't know the exact size of M3, but I assume we're talking about ~150mm2 or less.

For what it's worth Lunarlake's "compute" die also includes the GPU.

 

[coercitiv]

L3 was mentioned. "LLC", but only for the P.

Right, so that could explain the 8+8MB L3 reading from SiSoft Sandra, as both LLC and System Cache could be 8MB.

 

[DavidC1]

No kidding. When will Intel get their ...crap? together? I have zero faith in ARL at this point.

Meteorlake is underwhelming because circumstances made it delayed. Delays happen due to bad execution or not meeting targets, so they lengthen the cycle to try to mitigate it. There's never a "delay to make it better". You can't, because the product lifecycles are so short.

As an example, the "Knights Landing" Xeon Phi was supposed to come late 2015 with 200W TDP and 3TFLOP+ compute. It came out in middle of 2016.

Late 2015 with 3-3.2TFlop DPFP @ 200W vs
Mid 2016 with 3TFlop DPFP @ 240W

10% performance difference with nearly 20% lower power and 9 months gap. You can thank their 14nm for that.

If it wasn't delayed, it would have done better. To give you an idea how late Meteorlake was, Raptorlake replaced Meteorlake's original position.

I don't think we'll see any serious delays anymore. The bad actors are gone. Just like you saw the immediate slowing down of Apple's Mx development, it's all about the people working on it.

 

[FlameTail]

Nonsense, Intel just needs to pick employees with bigger hands to show the chips.

I don't know the exact size of M3, but I assume we're talking about ~150mm2 or less. IRRC this seems to be a size both Intel and AMD tried to hit in the past with their mobile dies, in the sense that they usually made sacrifices to stay around this mark (core count, cache etc). Therefore, for good or worse, this was likely the intended area target.

BTW, is there clear info on LNL cache structure? I remember seeing the diagram with 8MB "System Cache" but no L3 mentioned, and the Sisoft Sandra submission wasn't much help either.

Yes M3 is 146 mm².

 

[FlameTail]

So LNL basically has the same CPU cores as ARL, but on a better node (18A)?

That and Battlemage iGPU instead of Alchemist in ARL.

 

[Geddagod]

No kidding. When will Intel get their ...crap? together? I have zero faith in ARL at this point.

U can say crap without getting the ban hammer? Today I learned

Nonsense, Intel just needs to pick employees with bigger hands to show the chips.

I don't know the exact size of M3, but I assume we're talking about ~150mm2 or less. IRRC this seems to be a size both Intel and AMD tried to hit in the past with their mobile dies, in the sense that they usually made sacrifices to stay around this mark (core count, cache etc). Therefore, for good or worse, this was likely the intended area target.

BTW, is there clear info on LNL cache structure? I remember seeing the diagram with 8MB "System Cache" but no L3 mentioned, and the Sisoft Sandra submission wasn't much help either.

M3 is ~150 ye.
LNL cache structure:
LNC+ with increased L1 compared to LNC, 3MB of private L2 (split into 0.5MB of fast L2 and 2.5MB of slower L2), ? on L3, SLC

here's never a "delay to make it better". You can't, because the product lifecycles are so short.

Looks at SPR

So LNL basically has the same CPU cores as ARL, but on a better node (18A)?

Rumored to be slightly better LNC. Think RWC>RPC. Same node, ARL should have N3 variants, as would LNL. I wonder if LNL might end up being on N3E vs ARL prob being on N3B, or if LNL development started too early to potentially make that switch.

That and Battlemage iGPU instead of Alchemist in ARL.

Alchemist+. Doubled the L2, XMX, not bad.

 

[DrMrLordX]

LNL gets shown off. Compute die, SOC die, and rumored dummy structural die.
Area efficiency doesn't look too great TBH. The compute die alone on N3 looks to be the same size as the M3 die.

Which is the compute tile? Is it the one on the bottom right?

 

[mikk]

Which is the compute tile? Is it the one on the bottom right?

The bigger one is the compute tile (CPU+GPU).

 

[DrMrLordX]

The bigger one is the compute tile (CPU+GPU).

Oh okay. I figured that was the SoC tile.