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

Tigerick · Aug 22, 2022

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

igor_kavinski · Saturday at 12:41 PM

gdansk said:
Do you know anything about Pat?

Where is Battlemage dGPU if he's so obsessed with GPUs? He laid off members of the GPU driver team previously. How come it took until Arrow Lake to double the iGPU's performance? Why couldn't he release a better iGPU with 14th gen? How come Arrow Lake has an NPU not meeting Microsoft's Copilot+ requirements? Come on. Just admit that he's a SLOW decision maker. Too slow for a cutthroat business that Intel itself founded.

mzocyteae · Saturday at 12:43 PM

igor_kavinski said:
Sorry. Disagree.

Things he could've done:

1) Release Bartlett Lake type of CPU on latest refinement of Intel 7 instead of 14900KS (which was useless because it couldn't sustain high clocks without special cooling due to thermal throttling) with AVX-512 enabled.

2) If Meteor Lake was too much in demand (like he said), he should've tried to improve and fix Intel 4 yields before churning those out instead of incurring a huge loss with hot lots.

3) Intel had working Lunar Lake prototypes in October 2023 yet the laptops were ready for sale only a year later. Being a CPU architect, he should've understood the importance of Lunar Lake over Meteor Lake and sped up the development of the former no matter what. He probably didn't do that because MTL was on Intel 4 and he wanted Intel fabs to keep being utilized. That stupid plan backfired spectacularly on him.

4) He probably cut the funding of the GPU team and slowed down ARC's development to focus on other stuff he thought was more profitable. That was a bad move. We could've had Battlemage six months ago had he not messed with that team by firing a lot of them.

5) Gaudi 3? Pfffttt. How many billions did Intel make from THAT?

Pat's been too slow for Intel. Intel needs someone ambitious with a great sense of direction. Not some old has-been who can't figure out what Intel is good at.

These are not things that a people manager should do...
What he should do is to pick the good "professionals" and settle office politics, which he obviously failed to accomplish.

There is nothing wrong for 3), one year from prototype to product is pretty healthy.

gdansk · Saturday at 12:44 PM

igor_kavinski said:
Where is Battlemage dGPU if he's so obsessed with GPUs? He laid off members of the GPU driver team previously. How come it took until Arrow Lake to double the iGPU's performance? Why couldn't he release a better iGPU with 14th gen? How come Arrow Lake has an NPU not meeting Microsoft's Copilot+ requirements? Come on. Just admit that he's a SLOW decision maker. Too slow for a cutthroat business that Intel itself founded.

Because the reality of Intel's situation doesn't afford infinite agency. How many more tape outs would be needed to make the ARL of your dreams? So many that it would launch after Panther Lake. It's an enhanced Meteor Lake and there isn't much fixing that.

Kocicak · Saturday at 12:46 PM

maddie said:
This isn't true. Both threads in a HT core shares the resources. There isn't 1 strong and 1 weak one, or at least not continually. The threads are dynamic.

HT adding, say 30% to throughput does not mean 1 thread is at 100% and the other stays at 30%. It could be 65% + 65%, or any mix within the limits.

You have NO idea what are you talking about. There are three types of threads:

1st thread on P core,
thread on E core
2nd thread on P core.

These threads are utilised in this order, as the thread count of the load increases.

If you have 8 thread load, all these threads will be on P cores.

Each thread above 8 will be placed on E core.

Only after you run out of unutilised cores, the second threads on P cores will be utilised.

For example for 14700K, CPU with loads with 20 and less threads will perform exactly the same whether the HT is on or off. Only the extreme 28 thread loads will utilise HT and will improve performance of these loads by about 10%.

HT is not needed anymore for normal consumer workloads.

gdansk · Saturday at 12:48 PM

Kocicak said:
HT is not needed anymore for normal consumer workloads

It isn't needed. But it has been around for decades so consumer and server operating systems generally handle SMT competently. Meanwhile we still have scheduling problems every time there is a new heterogenous core design launching for Windows...

igor_kavinski · Saturday at 12:51 PM

511 said:
Show me the proof for this

New Intel Layoffs Impact GPU And Cloud Software Staff Among Wide Range Of Roles | CRN

Intel plans to lay off more than 300 California employees in a wide range of roles, including software development for GPUs, cloud computing and AI computing.

www.crn.com

The affected roles in those areas consisted of 11 GPU software development engineers and two graphics hardware engineers

Since it was a cost cutting move, you can bet those engineers were well paid and thus laying them off would help them save a lot of money. They wouldn't have hired them in the first place if they weren't required for the roles they were hired for. The GPU team had already seen cutbacks during the previous AXG restructuring.

cannedlake240 · Saturday at 12:53 PM

gdansk said:
Do you know anything about Pat? The only reason there are still plans for discrete GPUs is because he is obsessed with Intel having their own graphics that they can eventually scale into DC. Any other CEO would have killed all those plans by now.

He has said in multiple interviews that not having GPGPU in the 2010s was a huge miss by Intel, even before he was CEO.

Nah this is true, there's a reason Falcon shores has been completely redefined and celestial dgpu is rumored to be canned completely. There's also been multiple layoffs at their main GPU development site in 2023

gdansk · Saturday at 12:56 PM

cannedlake240 said:
Nah this is true, there's a reason Falcon shores has been completely redefined and celestial dgpu is rumored to be canned completely. There's also been multiple layoffs at their main GPU development site in 2023

I argued that any other CEO would have cut more and earlier. It seems like a weird thing to criticize Pat for unless you are adamant he should have done so earlier. He delayed most the GPU cuts until Intel posted a billion dollar loss.

cannedlake240 · Saturday at 1:02 PM

gdansk said:
I argued that any other CEO would have cut more and earlier. It seems like a weird thing to criticize Pat for unless you are adamant he should have done so earlier. He delayed most the GPU cuts until Intel posted a billion dollar loss.

Pat critics believe Intel should've gotten rid of the fabs and focused purely on the design side. Under his management Intel exited many businesses which of course affects their growth prospects that are potentially more profitable than maintaining a leading edge foundry

gdansk · Saturday at 1:04 PM

cannedlake240 said:
Pat critics believe Intel should've gotten rid of the fabs and focused purely on the design side. Under his management Intel exited many businesses which of course affects their growth prospects that are potentially more profitable than maintaining a leading edge foundry

Yeah... and that isn't what Igor said at all. He definitely is not cutting GPU to focus on things that are more profitable. He has a focus on less profitable (fabs) because he thinks it may have a long term future value for Intel. And GPU isn't being cut as much as it should be if discrete BM launches.

RTX2080 · Saturday at 1:31 PM

poke01 said:
https://twitter.com/x/status/1849838602920002007

Ouch. Translation: I have nothing to say... It's basically like Intel used 3nm to make Zen4.

Continuous take:

https://twitter.com/x/status/1850182596870996230

https://x.com/hjc4869/status/1850171375266439445

I wouldn't judge this is a Rocketlake-like launch, but this looks more like a terrible design choice. 3nm node advantage almost disappear on ARL. Don't tell me this is still going to happen on post ARL / future lakes.

jpiniero · Saturday at 1:40 PM

RTX2080 said:
I wouldn't judge this is a Rocketlake-like launch, but this looks more like a terrible design choice. 3nm node advantage almost disappear on ARL. Don't tell me this is still going to happen on post ARL / future lakes.

N3B isn't much better quality wise than N4P. You do get a decent logic shrink but that's about it.

511 · Saturday at 1:50 PM

igor_kavinski said:
New Intel Layoffs Impact GPU And Cloud Software Staff Among Wide Range Of Roles | CRN

Intel plans to lay off more than 300 California employees in a wide range of roles, including software development for GPUs, cloud computing and AI computing.

www.crn.com

Since it was a cost cutting move, you can bet those engineers were well paid and thus laying them off would help them save a lot of money. They wouldn't have hired them in the first place if they weren't required for the roles they were hired for. The GPU team had already seen cutbacks during the previous AXG restructuring.

That was like a year and a half ago also if someone is laid someone is hired back unless whatever they work on is Canned

LightningZ71 · Saturday at 1:59 PM

Saylick said:
Will be interesting to see how the convergence plays out. I have to imagine the size of the E core slowly grows until it basically becomes roughly the size of the competing Zen core, mostly because Intel’s P cores are bloated to begin with (AMD has been able to get similar levels of performance in a smaller die area) and E cores will need a lot more xtors to have feature parity with Zen, namely AVX512, but also clock parity.

Going forward, it looks like AMD will likely retain AVX-512 on client cores, but it'll look more like it does on the Strix Point P cores, double pumped 256 instead of full throughput 512 like on Granite Ridge. Remember, next gen is supposed to be the split where client and server begin to diverge into separate cores and desktop shares mobile cores/CCDs.

igor_kavinski · Saturday at 2:02 PM

511 said:
That was like a year and a half ago also if someone is laid someone is hired back unless whatever they work on is Canned

I don't have to inform you how HR works as you probably have a better idea about it. They are extremely impersonal and dispassionate and usually do what they are told by the higher ups without doing any digging into whether losing a particular employee would be bad for the company's future. It also means that if your manager wants to wreck your career due to some personal issue with you, HR won't stop them.

maddie · Saturday at 2:53 PM

igor_kavinski said:
That's true. However, ST performance of Raptor Lake improving a bit when HT is turned off (and quite a few gamers have been doing that since Alder Lake launch) means that some resources get statically partitioned so it's not dynamic HT. Turning on HT starves the primary thread of some resources it could've had if HT hadn't been turned on.

This is another issue, not strong thread and weak thread argument.

Meteor Late · Saturday at 3:10 PM

jpiniero said:
N3B isn't much better quality wise than N4P. You do get a decent logic shrink but that's about it.

Yet Lion Cove is still bigger than a Zen 5 core!
N3B is slightly better performance-power wise than N4P, at least in the low to mid curve. I've seen power numbers with Spec and yeah even at lower frequencies this thing is not good at all.

moinmoin · Saturday at 3:15 PM

poke01 said:
https://quasarzone.com/bbs/qf_cmr/views/2557513

Interview with David Huang.

these answers are interesting:

"The L3 latency of Luna Lake and Rapter Lake is about 55 cycles, while Arrow Lake (265K) is about 73 cycles and Meteo Lake (155H) is 80 cycles."

So this answers the previous question I had, Arrow Lake at least is an improvement of ~8% in L3$ latency. This makes Arrow Lake's L3$ 33% slower than Raptor Lake, an improvements over Meteor Lake's even worse 45%. But it's interesting to see that Lunar Lake is equal to Raptor Lake. Sounds like Intel either mismanaged the desktop market or just didn't care enough about it.

adroc_thurston · Saturday at 3:26 PM

moinmoin said:
But it's interesting to see that Lunar Lake is equal to Raptor Lake.

Way smaller ring.

moinmoin · Saturday at 4:03 PM

adroc_thurston said:
Way smaller ring.

So? AMD also started with CCX containing only 4 cores. Who says Intel couldn't have done something similar, scaling to desktop essentially using two or more Lunar Lakes. Instead they changed nothing and improved too little over Meteor Lake.

Hitman928 · Saturday at 4:07 PM

511 said:
That also Not getting the CEO role was the reasons he left if Pat was CEO i am pretty sure Intel would have been on a much better state than rn he kept saying Intel Needed GPUs but MBAs
He need go lay off some upper management

Funny story, before Pat left the CEO (Otellini) wanted to buy Nvidia because he saw the potential for a data center GPU. The board of directors pushed back on the cost to buy NV and instead, they were convinced by a team at Intel that it would be a waste of money because they were working on a project that would provide their own DC GPU. That team was working on project Larrabee and was headed by Pat Gelsinger. To this day, Pat believes that Larrabee would have revolutionized the market if it had never been cancelled (in reality it was a disaster) and that NV's current AI dominance is due to them being "extraordinarily lucky."

I share this to say that I have strong doubts that Intel would be in a better state than they are now if he was the CEO back then. I do think he's done some good things at Intel compared to the recent ex-CEOs, but I also see some of his moves that show the "old think" at Intel that led to so many of their problems to begin with.

AcrosTinus · Saturday at 4:14 PM

Ian Cutress said it the best in his tweets, Intel was never in position to satisfy the tech reviewer crowd.

What that crowed wanted:

A CPU from Intel that is faster than the AMD 9950X
Faster than the 7800x3D and potentially significantly faster so that the upcoming 9800x3D is within rage as well.
While consuming as much energy as the 7800x3D, a chip with half its guts missing and just a fraction of the MT performance.

You see how constrained the whole tech media space is, in any case Intel would get negative press.

Scenarios:

Intel goes for the performance again, with a higher base TDP, aggressive clocks and very high power draw. You get in all cases better performance than the 14900K
- Result: It would be compared to the 7800x3D, we would get 100s of videos with a flames in the thumbnail. No one would even acknowledge the performance, all would focus on the power draw in order to negate the competitive performance while ignoring the MT performance. The fate of the 13900K and 14900K would repeat itself.
Intel goes for a imaginary skew that uses the old Intel 7 and only 8 Lion Cove cores with HT enabled and your favorite fairy dust Level 4 Cache with 100MB. Let us say it just exceeds the 7800x3D by 5%.
- Result: The power draw, now of course MT performance would matter to reviewers, they will start benchmarking shader compilation.
- They will remind you in every step that the platform might never get a new CPU and that AMD will last until the heat death of the galaxy. (Intel always has 2 to 3 gens and never ever has confirmed in the past)
- They will start discouraging people from considering it by mentioning the 9800x3D that has not launched and plastering frame per watt graphs everywhere. There might be an additional disclaimer to tell people that even though this Intel 7 Gen3 is said to be safe every reviewer would tell people to not trust it and that it is a old 10NM node in reality that has failed already.
Intel goes for efficiency with no performance gain.
- Result: mention that the platform is dead, say that AMD is better due to AMD lasting until the end of time. Compare it to the 7800x3D again, then mention the 9800x3D and tell everyone to not buy it.
- DOOMSDAY:_
  - Go for efficiency with performance regression:
    - Everything mentioned above.
    - There was never a world where Intel would get positive media coverage but meeting the crazy standard mentioned at the start.
    - As they should because any imaginary world I mentioned is better than where we ended up.

adroc_thurston · Saturday at 4:15 PM

moinmoin said:
scaling to desktop essentially using two or more Lunar Lakes

well PTL-S was kinda that.

moinmoin said:
AMD also started with CCX containing only 4 cores.

That's not really fair, AMD LLC implementation is very very very good.

AcrosTinus said:
What the crowed wanted:

A CPU from Intel that is faster than the AMD 9950X

Faster than the 7800x3D and potentially significantly faster so that the upcoming 9800x3D is within rage as well.

While consuming as much energy as the 7800x3D, a chip with half its guts missing and just a fraction of the MT performance.

The crowd wanted something at least incrementally faster than RPL-S at lower power (it's a 2 node shrink ffs) and got neither.
Redundant cope.

gdansk · Saturday at 4:22 PM

AcrosTinus said:
Ian Cutress said it the best in his tweets, Intel was never in position to satisfy the tech reviewer crowd.

What that crowed wanted:

A CPU from Intel that is faster than the AMD 9950X

Faster than the 7800x3D and potentially significantly faster so that the upcoming 9800x3D is within rage as well.

While consuming as much energy as the 7800x3D, a chip with half its guts missing and just a fraction of the MT performance.

You see how constrained the whole tech media space is, in any case Intel would get negative press.

Scenarios:

Intel goes for the performance again, with a higher base TDP, aggressive clocks and very high power draw. You get in all cases better performance than the 14900K

Result: It would be compared to the 7800x3D, we would get 100s of videos with a flames in the thumbnail. No one would even acknowledge the performance, all would focus on the power draw in order to negate the competitive performance while ignoring the MT performance. The fate of the 13900K and 14900K would repeat itself.

Intel goes for a imaginary skew that uses the old Intel 7 and only 8 Lion Cove cores with HT enabled and your favorite fairy dust Level 4 Cache with 100MB. Let us say it just exceeds the 7800x3D by 5%.

Result: The power draw, now of course MT performance would matter to reviewers, they will start benchmarking shader compilation.

They will remind you in every step that the platform might never get a new CPU and that AMD will last until the heat death of the galaxy. (Intel always has 2 to 3 gens and never ever has confirmed in the past)

They will start discouraging people from considering it by mentioning the 9800x3D that has not launched and plastering frame per watt graphs everywhere. There might be an additional disclaimer to tell people that even though this Intel 7 Gen3 is said to be safe every reviewer would tell people to not trust it and that it is a old 10NM node in reality that has failed already.

Intel goes for efficiency with no performance gain.

Result: mention that the platform is dead, say that AMD is better due to AMD lasting until the end of time. Compare it to the 7800x3D again, then mention the 9800x3D and tell everyone to not buy it.

DOOMSDAY:_

Go for efficiency with performance regression:

Everything mentioned above.

There was never a world where Intel would get positive media coverage but meeting the crazy standard mentioned at the start.

Intel has P and E cores. Intel has a node density advantage. Intel has a packaging advantage. Arrow Lake ends up with same die space as 2x CCD Granite Ridge. It should be easy for them to have something better than 9950X in some way. But it lost to regular Zen 5 in gaming and only matched the 7950X in Phoronix. It should have won one or the other to have appeal but it lost both...

AcrosTinus · Saturday at 4:23 PM

adroc_thurston said:
well PTL-S was kinda that.

That's not really fair, AMD LLC implementation is very very very good.

The crowd wanted something at least incrementally faster than RPL-S at lower power (it's a 2 node shrink ffs) and got neither.
Redundant cope.

Yes I am coping, I'm losing real money here

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

Senior member

Attachments

Lifer

Junior Member

Diamond Member

Golden Member

Diamond Member

Lifer

Member

Diamond Member

Member

Diamond Member

Senior member

Lifer

Senior member

Golden Member

Lifer

Diamond Member

Member

Diamond Member

Diamond Member

Diamond Member

Diamond Member

Member

Diamond Member

Diamond Member

Member