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.)

 

[poke01]

Though Meteor Lake certainly isn't likely to do Intel's reputation any favors, regardless of the health of Intel 4.

In your view what would Intel have to do regain rep. Beat TSMC in nodes, packaging? Release Lunar Lake and Royal core that brings leadership in perf/w. Server is already a lost cause for Intel until 2028(Let's be real).

 

[igor_kavinski]

Server is already a lost cause for Intel until 2028(Let's be real).

They should be fine if they can actually release Sapphire Rapids. A lot of workloads need the specialized accelerators they are promising. Their mistake was trying to chew on way more than they could fit in their mouths. They could have released a plain Alder Lake server variant ASAP to start clawing back their server marketshare. But I guess they have some really headstrong, dumb project leaders there.

 

[jpiniero]

Their highest volume products are Core i3 class parts. For example back in 2017, the dual core 100mm2 chips were the majority of sales. Yes, Core is highest volume client category, not Pentium/Celeron since a decade ago, but you are still talking about the lower end.

No, it's i5. By a wide margin too. i3 is still however very popular.

Core Pentium/Celeron is a tiny part of the market.

 

[igor_kavinski]

No, it's i5.

You guys are both arguing without any proof about your claims.

 

[jpiniero]

You guys are both arguing without any proof about your claims.

I don't have any recent info, but just in my experience the cheapest i5 in particular is by far the most popular. OEMs will try to upsell people to i7 but most companies won't bother.

If you look at Dell's website, they aren't even selling any 12th Gen i3 laptops right now. Part of that might be because they are preferring to sell Tiger Lake i5 instead.

 

[igor_kavinski]

If you look at Dell's website, they aren't even selling any 12th Gen i3 laptops right now. Part of that might be because they are preferring to sell Tiger Lake i5 instead.

On the AMD side, I can believe that because I see a LOT of 5500U laptops in my usual laptop searches on Amazon. But at the same time, there is no dearth of Core i3-1115G4 laptops.

 

[Exist50]

In your view what would Intel have to do regain rep. Beat TSMC in nodes, packaging? Release Lunar Lake and Royal core that brings leadership in perf/w. Server is already a lost cause for Intel until 2028(Let's be real).

For the "perma bears", I don't think there's any realistic scenario that would change their minds. But for the broader audience, what Intel has to do is very simple - release competitive products when they said they would. So in some sense, all of the above. Though just fyi, Lunar Lake isn't Royal.

 

[Geddagod]

I think what actually constitutes as 'Royal Core' is hotly debated.
MLID (who also originally leaked Royal Core Project to my knowledge) makes it seem like Royal Core is more of a series of architecture or design choices, where the following architectures after redwood cove would use bits and parts of.
He claims the architecture in Arrow Lake is part of the 'royal core' project.

 

[Exist50]

I think what actually constitutes as 'Royal Core' is hotly debated.
MLID (who also originally leaked Royal Core Project to my knowledge) makes it seem like Royal Core is more of a series of architecture or design choices, where the following architectures after redwood cove would use bits and parts of.
He claims the architecture in Arrow Lake is part of the 'royal core' project.

The problem there is assuming that MLID has the faintest clue what he's talking about. So in no uncertain terms, no, he's simply wrong.

 

[IntelUser2000]

Where did you hear most of the volume is i3s?

In 2011 client group had revenue of $8-9 billion per quarter. So it's about the same as today. If i5 is the major lineup now, then it would just mean they degraded their product value to get that claim.

There was a detailed breakdown of Intel ASPs per product line I saw and their ASP was $100-110. It wasn't until by going firmly into the i5 line that the ASPs started really rising from there. The top i7 was something like $200.

Think about what you are saying here. Their lineup consists mostly of Core i5's with $140-150 ASPs but even with the chipset, the company ASP is only $100(Q2 data)? That either means there's a LOT of Pentium/Celeron sales below the $100 mark or the most sold lineup is really low end of the Core i3 line.

Yes, Dell has quite a few models too, but Core i5 ones are easily in the $599 range and they don't deal with lot of offline sales. Think about the Chromebooks, and the no name "HP 15" devices. The results might be even more skewed in other places of the world where the products are marked up quite a bit higher than currency conversion rates suggest(even here in Canada).

 

[jpiniero]

In 2011 client group had revenue of $8-9 billion per quarter. So it's about the same as today. If i5 is the major lineup now, then it would just mean they degraded their product value to get that claim.

There was a detailed breakdown of Intel ASPs per product line I saw and their ASP was $100-110. It wasn't until by going firmly into the i5 line that the ASPs started really rising from there. The top i7 was something like $200.

Think about what you are saying here. Their lineup consists mostly of Core i5's with $140-150 ASPs but even with the chipset, the company ASP is only $100(Q2 data)? That either means there's a LOT of Pentium/Celeron sales below the $100 mark or the most sold lineup is really low end of the Core i3 line.

Don't confuse what you pay at Newegg with what Intel actually recieves in revenue.

 

[DrMrLordX]

The problem there is assuming that MLID has the faintest clue what he's talking about. So in no uncertain terms, no, he's simply wrong.

That is one problem. The other is in Royal Core not becoming another Ocean Cove.

 

[Exist50]

That is one problem. The other is in Royal Core not becoming another Ocean Cove.

Nah, not at this point. At least not under Gelsinger. We'll see Royal eventually. Just a question of what and when.

 

[poke01]

Nah, not at this point. At least not under Gelsinger. We'll see Royal eventually. Just a question of what and when.

Probably after Lunar Lake. But Intel says Lunar Lake will have pref/w leadership. So Intel has to beat Apple, Qualcomm and AMD in pref/w.

AMD could be easy to beat but Apple and Qualcomm invest so much more in efficiency. It will be very hard to beat these two. Of cource I am talking about laptop/mobile, this area will be huge in the future. Almost everybody needs a laptop/tablet but not everyone needs a desktop. So having excellent mobile SoC's is very important. In anycase I will take Intel's claims as marketing unless proven.

 

[Exist50]

Probably after Lunar Lake. But Intel says Lunar Lake will have pref/w leadership. So Intel has to beat Apple, Qualcomm and AMD in pref/w.

AMD could be easy to beat but Apple and Qualcomm invest so much more in efficiency. It will be very hard to beat these two. Of cource I am talking about laptop/mobile, this area will be huge in the future. Almost everybody needs a laptop/tablet but not everyone needs a desktop. So having excellent mobile SoC's is very important. In anycase I will take Intel's claims as marketing unless proven.

Even with the usual caveats about marketing, that byline was under "Lunar Lake & Beyond", so who knows what it actually translates to. But yeah, Lunar Lake doesn't have Royal.

 

[IntelUser2000]

AMD could be easy to b
beat but Apple and Qualcomm invest so much more in efficiency.

We know it's possible since Intel has achieved it back with the Tablet Atoms. The Pentium Silver chips after that didn't achieve the efficiency the Tablet Atoms did.

Part of the loss is the cost cutting on the laptop side to get to real low price points. Other is the loss of efficiency since Cometlake. Ice/Tiger/Alder all in one form or another lowered battery life compared to predecessors.

 

[Hulk]

In your view what would Intel have to do regain rep. Beat TSMC in nodes, packaging? Release Lunar Lake and Royal core that brings leadership in perf/w. Server is already a lost cause for Intel until 2028(Let's be real).

If AMD (and TMSC) hadn't been able to execute Zen as they have Intel wouldn't have appeared to be having troubles, or they could have hid them better if they were able to maintain market dominance with Skylake on their schedule, as I think they had planned. My point is that for me it's not to much Intel's failure to execute but AMD's ability to capitalize on their failures.

To summarize, in my mind Intel is still Intel, it's AMD that has changed.

 

[dangerman1337]

Has there been any hints last few weeks or last month of when MTL-S hits? I hope it's Summer-ish next year because the £ taken a crap in the UK and the prices of Motherboards + CPUs + no signs of Hynix A Die DDR5 hitting this year in the UK + no White 1000+W ATX 3.0 PSUs despite Asus prepping them. This 1080 Ti + 8700K + SATA SSD feeling old at 2560x1080 @ 144Hz.

 

[jpiniero]

Has there been any hints last few weeks or last month of when MTL-S hits? I hope it's Summer-ish next year because the £ taken a crap in the UK and the prices of Motherboards + CPUs + no signs of Hynix A Die DDR5 hitting this year in the UK + no White 1000+W ATX 3.0 PSUs despite Asus prepping them. This 1080 Ti + 8700K + SATA SSD feeling old at 2560x1080 @ 144Hz.

Who knows but it's extremely unlikely that it's any sooner than a year from Raptor Lake's launch... which is 9 days from now.There's also no guarantee that it will even make it to market or only be low end and Intel just does a rebrand of Raptor Lake for 14th Gen so they have something 'new' and calls it a day.

 

[Saylick]

Has there been any hints last few weeks or last month of when MTL-S hits? I hope it's Summer-ish next year because the £ taken a crap in the UK and the prices of Motherboards + CPUs + no signs of Hynix A Die DDR5 hitting this year in the UK + no White 1000+W ATX 3.0 PSUs despite Asus prepping them. This 1080 Ti + 8700K + SATA SSD feeling old at 2560x1080 @ 144Hz.

Rumors say Q4 2023...

 

[DrMrLordX]

Rumors say Q4 2023...

The big problem I see for Meteor Lake going forward is non-technical in nature: the current economic climate is making it less-desirable for orgs to spend on PCs/notebooks. Meteor Lake should be a nice improvement over Alder/Raptor mobile (well 6+8 configs anyway), yet orgs are far less likely to adopt it. If Intel sits on Meteor Lake long enough they may avoid the worst of the recession. Smart money's on orgs doing mass upgrades approaching the Win10 end-of-support deadline in 2025 though.

 

[IntelUser2000]

My point is that for me it's not to much Intel's failure to execute but AMD's ability to capitalize on their failures.

It's both. Even AMD is behind, because they had to make up for their own mistakes in the past.

Remember the two x86 vendors are "notorious" for having many delays and missteps in their history. You can see by how the ARM vendors do so much better.

And we all know that Intel failed badly because prior to 2013 they were undisputed king of process technology, because they were able to marry process and uarch very well. They stumbled so long that they took the 2 year lead and turned it into a 2 year loss.

You can also say since engineers move from one company to other, in a sense Intel's failures directly contributed to AMD and others doing better.

 

[poke01]

You can also say since engineers move from one company to other, in a sense Intel's failures directly contributed to AMD and others doing better.

But Intel management was aslo bad.

It's both. Even AMD is behind, because they had to make up for their own mistakes in the past.

Remember the two x86 vendors are "notorious" for having many delays and missteps in their history. You can see by how the ARM vendors do so much better.

And we all know that Intel failed badly because prior to 2013 they were undisputed king of process technology, because they were able to marry process and uarch very well. They stumbled so long that they took the 2 year lead and turned it into a 2 year loss.

You can also say since engineers move from one company to other, in a sense Intel's failures directly contributed to AMD and others doing better.

I know Nvidia has bad ethics but man their lineup has almost never been weak arch wise.

 

[jpiniero]

gcc.gnu.org Git - gcc.git/blobdiff - gcc/common/config/i386/i386-common.cc

gcc.gnu.org

In case there was any doubt, no AVX-512 on Meteor Lake.

 

[moinmoin]

gcc.gnu.org Git - gcc.git/blobdiff - gcc/common/config/i386/i386-common.cc

gcc.gnu.org

In case there was any doubt, no AVX-512 on Meteor Lake.

Considering how Apple Silicon and ARM SoC keep being extended with accelerators and AMD positions AVX-512 and Xilinx's AIE for that purpose it will be really interesting how Intel intends to move forward. Stagnation is not really an option, especially in an environment with strong competition.