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.





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.)

 

[SiliconFly]

Anyone remembers this MLID "leak"?

Intel 14th Gen Core 'Meteor Lake': 15-21% IPC boost over Raptor Lake

Intel's new 14th Gen Core 'Meteor Lake' CPUs rumored to have 15-21% more IPC performance over 13th Gen Core 'Raptor Lake' CPUs.

www.tweaktown.com

"
We are being teased with 15-21% increases in IPC performance with the new Redwood Cove architecture over the Raptor Cove architecture, but with MLID apologizing for the huge range (15-21%) but I think that's fine."

No biggie, MLID missed it will be 0-1% , just a 100% failed prediction. I posted this here just to show how this guy either :
1) has no real sources
2) is making it all up for clicks and
3) has multiple "sources" who are actually trolls and who feed him BS info all the time

Very true. He's a dumbass.

 

[Mopetar]

• They expect MTL iGPU to be the best performing iGPU in the market.

That's certainly an interesting claim and even if remotely true means Intel has seen some massive gains in their GPU technology.

Here's a few charts from AT iGPU tests of Raptor Lake:

There are some games where they're comparable, or even ahead, but others where they just get smoked so hard it's difficult to believe they could accomplish that

Even still, I want to believe it's loss just so they can shake up the GPU market even more with that technology. The GPU market needs a good shaking.

 

[SiliconFly]

He receive some noise but it look like he has trouble making the difference between IPC, frequency, better perf/watt that allow MT frequencies uplifts and so on, otherwise his 20% number is not far from other leaks, he s just unable to point to what parameter it is exactly related.

He once mentioned that he isn't an engineer and doesn't understand technical stuff. And based on his silly MTL predictions & ridiculous ARL projections, it's clear he doesn't understand a thing about foundry node characteristics. With that level of ignorance, he shouldn't be making bold predictions without proper understanding. It's just wrong.

 

[Hulk]

He once mentioned that he isn't an engineer and doesn't understand technical stuff. And based on his silly MTL predictions & ridiculous ARL projections, it's clear he doesn't understand a thing about foundry node characteristics. With that level of ignorance, he shouldn't be making bold predictions without proper understanding. It's just wrong.

It's pure entertainment.

 

[Abwx]

That's certainly an interesting claim and even if remotely true means Intel has seen some massive gains in their GPU technology.

Here's a few charts from AT iGPU tests of Raptor Lake:

Even still, I want to believe it's loss just so they can shake up the GPU market even more with that technology. The GPU market needs a good shaking.

The games where they are comparable are those where CPU troughput matter, the others are GPU bound...

 

[SiliconFly]

That's certainly an interesting claim and even if remotely true means Intel has seen some massive gains in their GPU technology.

Here's a few charts from AT iGPU tests of Raptor Lake:

There are some games where they're comparable, or even ahead, but others where they just get smoked so hard it's difficult to believe they could accomplish that

Even still, I want to believe it's loss just so they can shake up the GPU market even more with that technology. The GPU market needs a good shaking.

Intel themselves claim that the gains maybe be up to 2X. But we need to take it with a grain of salt, as the game performance is not only dependent on gpu, but also cpu, memory, bandwidth, etc. They may be talking about best case scenarios.

 

[DrMrLordX]

What if it's two compute tiles since they are claiming that power efficiency has increased by 2X? 12P+16E wouldn't be bad. At least, not horribad.

That product has been kept well under wraps if Intel has any plans of releasing it.

 

[trivik12]

MJ is saying MTL desktop in 2024.

https://twitter.com/x/status/1705676919851925582

Also confirmed by MTL design head. 1st time design was more complex but going forward they can scale it for all power envelopes.

PCMAG: Given that all the discussions have been about laptops and mobile, is there anything intrinsic that Meteor Lake might mean for future desktop designs as distinct from mobile?

WILSON: Good question, good question. That's top of mind for a lot of people. I'll just say up front, this architecture will scale top to bottom in our client segment. Desktop all the way down to mobile. With Meteor Lake, we're launching mobile first. And that largely has to do with, if you think back to the keynote that I gave, the four key design principles around which we design the product, those really go toward the mobile segments first. Our most power efficient SoC we ever built...we want that in mobile. That 2x integrated graphics capability—bringing discrete into an integrated form factor—that goes to mobile. AI applies to both equally. I would say, but because of the value proposition of the product, what we were trying to build, we targeted mobile first. But you will see this architecture scale top to bottom—up to desktop, as well.

 

[cebri1]

I’m betting on a limited released and only available on pre-built desktops. For small NUCs does make sense, low power + discrete graphics. Nothing for DIY.

 

[Mopetar]

The games where they are comparable are those where CPU troughput matter, the others are GPU bound...

I just pulled out a few of the different AT results just to give people a picture of how far they have to go in some cases.

Intel themselves claim that the gains maybe be up to 2X. But we need to take it with a grain of salt, as the game performance is not only dependent on gpu, but also cpu, memory, bandwidth, etc.

2X is rather big jump. Have they given any indication of how they intend to get there? It's certainly possible to do just be throwing more resources at the problem, but that's where the bandwidth, etc. limits start to rear their head.

I was mostly hoping that they'd made some major improvements in their GPU tech, because that will mean their next set of GPUs should also see similar gains even without adding more cores.

 

[mikk]

Anyone remembers this MLID "leak"?

Intel 14th Gen Core 'Meteor Lake': 15-21% IPC boost over Raptor Lake

Intel's new 14th Gen Core 'Meteor Lake' CPUs rumored to have 15-21% more IPC performance over 13th Gen Core 'Raptor Lake' CPUs.

www.tweaktown.com

"
We are being teased with 15-21% increases in IPC performance with the new Redwood Cove architecture over the Raptor Cove architecture, but with MLID apologizing for the huge range (15-21%) but I think that's fine."

No biggie, MLID missed it will be 0-1% , just a 100% failed prediction. I posted this here just to show how this guy either :
1) has no real sources
2) is making it all up for clicks and
3) has multiple "sources" who are actually trolls and who feed him BS info all the time

I wouldn't say no sources, he definitely has there is no doubt. I can see this from the slides he is sharing sometimes and some infos with the code names, features, or even specs are just right. He was right with his 288 cores Sierra Forest forecast when everyone expected 144 cores. He is mixing real infos with false infos. His MTL IPC forecast was unrealistic from the beginning, it was crazy. Redwood Cove in Granite Rapids seems to differ with an enhanced uarch including improved branch predictor etc., it's like Redwood Cove+, exactly what MLID said. He said it comes with Redwood+ architecture on Intel 3 which gets 5-10% more IPC. I guess this is what Gelsinger meant when he said they enhanced Granite Rapids with new process and enhanced architecture: https://forums.anandtech.com/thread...-rapids-thread.2509080/page-614#post-40717471

 

[mikk]

That's certainly an interesting claim and even if remotely true means Intel has seen some massive gains in their GPU technology.

Here's a few charts from AT iGPU tests of Raptor Lake:

There are some games where they're comparable, or even ahead, but others where they just get smoked so hard it's difficult to believe they could accomplish that

Even still, I want to believe it's loss just so they can shake up the GPU market even more with that technology. The GPU market needs a good shaking.

Are you aware that the desktop chips only features 32 EUs compared to the mobile chips with 96 EUs? Also AMD didn't release a new desktop lineup yet with RDNA graphics expect for the 2CU version in Ryzen 7000 which is even slower than Intels GT1 Xe LP. It matters much more for mobile.

On paper Xe LPG is twice as fast as Xe LP due to a much higher clock speed and 33% more shader cores.

 

[Khato]

2X is rather big jump. Have they given any indication of how they intend to get there? It's certainly possible to do just be throwing more resources at the problem, but that's where the bandwidth, etc. limits start to rear their head.

I was mostly hoping that they'd made some major improvements in their GPU tech, because that will mean their next set of GPUs should also see similar gains even without adding more cores.

You can already compare the current Iris Xe against the Arc A370M to see how feasible a 2x performance increase is. The 112GB/s of the A370M isn't that far off the 120GB/s of shared bandwidth that the MTL pictured with on-package lpddr5x would have available.

 

[Geddagod]

I wouldn't say no sources, he definitely has there is no doubt.

Yes. Unfortunately, other than stuff that literally can not be misinterpreted wrong- like codenames- he gets nearly everything else wrong. Sigh

He was right with his 288 cores Sierra Forest forecast when everyone expected 144 cores.

You forgot the part where in a later video he backtracks and says the 288 core version got canned to focus on the 144 core version.

Redwood Cove in Granite Rapids seems to differ with an enhanced uarch including improved branch predictor etc., it's like Redwood Cove+, exactly what MLID said.

No it doesn't. Intel outright said that GNR uses redwood cove, not redwood cove+. Also MLID thought RWC+ had, 12-21% IPC over RPC (false). He maintained this even after he brought down his IPC estimates for the original RWC.

 

[Hulk]

Of course Meteor Lake is coming to the desktop in 2024. But it's gonna get a new name... Arrow Lake.

 

[DrMrLordX]

Of course Meteor Lake is coming to the desktop in 2024. But it's gonna get a new name... Arrow Lake.

Ugh let's hope that isn't the case.

 

[mikk]

No it doesn't. Intel outright said that GNR uses redwood cove, not redwood cove+. Also MLID thought RWC+ had, 12-21% IPC over RPC (false). He maintained this even after he brought down his IPC estimates for the original RWC.

It uses Redwood Cove, whether or not they officially call it Redwood Cove+ is not the point. The point is they are gonna use an updated/enhanced Redwood Cove over the Redwood Cove in Meteor Lake. Just compare the changes in Granite Rapids with Meteor Lake. It's enhanced in Granite Rapids whereas in Meteor Lake basically nothing, no IPC improvements in the core. Not even the usual improved branch predictor.

Gelsinger himself said they are using a major new core with 10 plus percent in the core. Redwood Cove in Meteor Lake certainly isn't major. I believe they planned with Intel 4 and Redwood Cove from Meteor Lake and later switched to Intel 3 and an enhanced Redwood Cove.

We did change the timing of Granite Rapids, and we had a big internal debate show on should we even keep the Granite Rapids name because it was the same platform, but it was a new core on a new process. So to some degree, it was a very different product. But some said, hey, you delayed Granite Rapids. Hey, I say I enhance Granite Rapids, with a much higher performance product, a much -- 18% process, a major new core, that's 10-plus percent in the core. So a much better product and aligned to the customers' timing.

 

[Geddagod]

The point is they are gonna use an updated/enhanced Redwood Cove over the Redwood Cove in Meteor Lake

No they won't. Other than perhaps the usual server upgrades (AMX, etc etc)

It uses Redwood Cove, whether or not they officially call it Redwood Cove+ is not the point.

It definitely is the point, esp considering the claim was 12-21% increase in ipc lol

It's enhanced in Granite Rapids whereas in Meteor Lake basically nothing, no IPC improvements in the core.

RWC in GNR is not going to have any IPC improvements over the RWC in MTL other than the usual server upgrades a core has. Also, prob the biggest change in RWC over Golden Cove was the change in L1I... which is seen in both diagrams.

Not even the usual improved branch predictor.

Just because it's not listed out in the slide doesn't mean it won't be there. I expect the full list of changes to be provided at MTL launch.

Gelsinger himself said they are using a major new core with 10 plus percent in the core. Redwood Cove in Meteor Lake certainly isn't major.

Ye, I think the consensus on what happened there is that they thought they would be able to cram in LNC into GNR, but wasn't able to get it done. At least that's what Exist50 thinks (and what I've been speculating about since GNR was switched to Intel 3).
Either that, or that 10% plus in the core was in reference to the total perf/watt uplift as a result of the "major new core" and the better process for Intel 4 GNR vs Intel 3 GNR.

Either way though, if you think the most important claim of my original paragraph was that MLID is wrong because he called RWC in GNR "RWC+" rather than just "RWC", you must not have read literally the rest of my comment lol. And if you did, and do seriously believe RWC in GNR is going to have a 12-21% IPC uplift over RWC in MTL, then idk what to tell you... ig you're just way more optimistic for Intel then me lol.

(P.S. check out my MLID Intel leak tracker to find out just how bad MLID is at leaking info

 

[mikk]

No they won't. Other than perhaps the usual server upgrades (AMX, etc etc)

Ok you have to prove this, can you? You are just saying NO without a real argument or proof from your side.

RWC in GNR is not going to have any IPC improvements over the RWC in MTL other than the usual server upgrades a core has.

Here again, you have to prove your claim. I have to remind you your track record isn't great either, remember your High-NA babble. It was so obvious or even confimed they won't use it for 18A and you still refused to believe it.

Just because it's not listed out in the slide doesn't mean it won't be there. I expect the full list of changes to be provided at MTL launch.

It isn't strange to you they call it enhanced in the Granite Rapids slide and didn't in the Meteor Lake slide? Enhanced with zero IPC improvements, this is realistic right? It isn't strange to you Gelsinger refers to an upgraded core with more performance and even considered a name change?

I think the consensus on what happened there is that they thought they would be able to cram in LNC into GNR, but wasn't able to get it done.

I don't think so. They upgraded the core and then downgraded it again, there is no sign this is what happened. In mid 2022 when Gelsinger talked about the upgraded core it wasn't realistic in hindsight to even consider a server variant of Lion Cove in 2024. And Lion Cove on Intel 3, I don't think so. We weren' aware there is an enhanced version of Redwood Cove and simply assumed it must be Lion Cove which shortly later turned out this wasn't the case. With the enhanced Redwood Cove it makes sense now.

 

[Geddagod]

Ok you have to prove this, can you? You are just saying NO without a real argument or proof from your side.

It's pretty simple, really. Intel called the core in MTL redwood cove. They called the core in Granite Rapids... redwood cove. Anyone claiming there's going to be any large changes between the two cores has the burden of proof to say they are different.

I have to remind you your track record isn't great either, remember your High-NA babble.

Because Intel changed their plans, my record (which btw I'm not a leaker on insider, I have made that clear numerous times) is bad? The only difference between me and you is that you choose to believe MLID (who I have empirically proven is a pretty terrible leaker).
Oh ye, and it was such babble that people in the industry also were speculating about High NA and 18A? Quoting Ryan Smith (an author for Anandtech)

Given that the release date of ASML’s High-NA machines has not changed, however, that announcement from Intel left open some questions about how High-NA would fit into their 18A node.

Btw, Intel announcing that high NA won't be used for 18a means that their claim of High NA EUV in HVM by 2025 was also a lie lol

or even confimed they won't use it for 18A

That's just a lie. They didn't confirm it until like 3 days ago.

It isn't strange to you they call it enhanced in the Granite Rapids slide and didn't in the Meteor Lake slide?

They called the core in MTL "new". This is the most marketing speak diction battle ever ffs

Enhanced with zero IPC improvements, this is realistic right?

Yes, because it's enhanced in different aspects - like efficiency. Oh, and I'm sure you could cherry pick some benches for RWC in both MTL and GNR to have higher IPC than GLC,.

It isn't strange to you Gelsinger refers to an upgraded core with more performance and even considered a name change?

Weird how that one piece of info is never mentioned again lol. You would think it would be a really neat piece of info to put up in Hotchips...

In mid 2022 when Gelsinger talked about the upgraded core it wasn't realistic in hindsight to even consider a server variant of Lion Cove in 2024.

Launch would be ~2 years away. That's cutting it slightly close, but GNR should still be in the design phase at that point, not after tape in. It's unlikely, but not impossible. And we have seen Intel work on the core or "redefine" products before as they get delayed. But that's an arguement you might want to have with Exist50, though he might be harder to find now (or just DM him on reddit lol).

And Lion Cove on Intel 3, I don't think so.

LNC on Intel 3 would be what is expected of Intel. RWC is ported GLC on a new node, LNC should be on Intel 3 as it's "mature Intel 4". It's Intel usual cadence (well what it should be anyway).

We weren' aware there is an enhanced version of Redwood Cove and simply assumed it must be Lion Cove

No one assumed that but me. And I made it clear plenty of times it was my ambitious speculation. Most people believed it to be RWC.

With the enhanced Redwood Cove it makes sense now.

No, because Intel claims GNR uses RWC. Do you really believe a standard P-core IPC upgrade is going to keep the same name as it's predecessor? Imagine calling Sunny Cove "skylake +" or Golden Cove "sunny cove+". Oh, and actually scratch that, it's even worse, Intel doesn't even call it +.
And that's if the interpretation that Gelsinger was referring to 10%+ as IPC, and not just a general perf/watt improvement as a result of the better node and improved architecture- which would also make sense.

 

[SiliconFly]

Maybe I shouldn't ask. But, do we really have to argue about a leaker whose predictions are based on his limited understanding of tech?

Info leaks are fine (whether it's real or fake). Keeps it interesting. But discussing the predictions of a baboon who lacks basic technical knowledge? Oh man!

 

[SiliconFly]

2X is rather big jump

WILSON: ...That 2x integrated graphics capability...

And not just the design head, I've seen two other videos where Intel people make similar claims. But like i mentioned, we have to take it with a pinch of salt. I think what they really mean is, that the iGPU's processing power has doubled. But that doesn't mean it'll directly translate to double the frame rates due to other obvious limitations mentioned previously.

 

[AMDK11]

It's pretty simple, really. Intel called the core in MTL redwood cove. They called the core in Granite Rapids... redwood cove. Anyone claiming there's going to be any large changes between the two cores has the burden of proof to say they are different.

Because Intel changed their plans, my record (which btw I'm not a leaker on insider, I have made that clear numerous times) is bad? The only difference between me and you is that you choose to believe MLID (who I have empirically proven is a pretty terrible leaker).
Oh ye, and it was such babble that people in the industry also were speculating about High NA and 18A? Quoting Ryan Smith (an author for Anandtech)

Btw, Intel announcing that high NA won't be used for 18a means that their claim of High NA EUV in HVM by 2025 was also a lie lol

That's just a lie. They didn't confirm it until like 3 days ago.

They called the core in MTL "new". This is the most marketing speak diction battle ever ffs

Yes, because it's enhanced in different aspects - like efficiency. Oh, and I'm sure you could cherry pick some benches for RWC in both MTL and GNR to have higher IPC than GLC,.

Weird how that one piece of info is never mentioned again lol. You would think it would be a really neat piece of info to put up in Hotchips...

Launch would be ~2 years away. That's cutting it slightly close, but GNR should still be in the design phase at that point, not after tape in. It's unlikely, but not impossible. And we have seen Intel work on the core or "redefine" products before as they get delayed. But that's an arguement you might want to have with Exist50, though he might be harder to find now (or just DM him on reddit lol).

LNC on Intel 3 would be what is expected of Intel. RWC is ported GLC on a new node, LNC should be on Intel 3 as it's "mature Intel 4". It's Intel usual cadence (well what it should be anyway).

No one assumed that but me. And I made it clear plenty of times it was my ambitious speculation. Most people believed it to be RWC.

No, because Intel claims GNR uses RWC. Do you really believe a standard P-core IPC upgrade is going to keep the same name as it's predecessor? Imagine calling Sunny Cove "skylake +" or Golden Cove "sunny cove+". Oh, and actually scratch that, it's even worse, Intel doesn't even call it +.
And that's if the interpretation that Gelsinger was referring to 10%+ as IPC, and not just a general perf/watt improvement as a result of the better node and improved architecture- which would also make sense.

I am of the same opinion that the same RedwoodCove(x86) is in GraniteRapids and MeteorLake.The difference in the presentation is that in the case of GraniteRapids Intel announced changes in the microarchitecture, and in the case of MeteorLake only that it is a new P Core.I believe that since Intel calls RedwoodCove a new core, it means that it is a generational jump and I assume an IPC jump in the range of 8-15%.

I am sure that more details, including numbers, will be provided on or before December 14.

 

[inf64]

I am of the same opinion that the same RedwoodCove(x86) is in GraniteRapids and MeteorLake.The difference in the presentation is that in the case of GraniteRapids Intel announced changes in the microarchitecture, and in the case of MeteorLake only that it is a new P Core.I believe that since Intel calls RedwoodCove a new core, it means that it is a generational jump and I assume an IPC jump in the range of 8-15%.

I am sure that more details, including numbers, will be provided on or before December 14.

Is there a chance that GraniteRapids uses ArrowLake cores? It would make sense that this is where the 10+% IPC increase comes from, since MeteorLake obviously brings close to zero IPC versus current gen parts.

 

[SiliconFly]

brings close to zero IPC versus current gen parts.

Actually, there's more to it. If we compare the leaked benchmarks of Core Ultra 7 with the equivalent i7-13700H (both operating at similar clocks), there's a slight performance regression. Possibly due to the introduction of the NOC interconnect fabric. Or maybe because the MTL samples are all ES.