Discussion Intel current and future Lakes & Rapids thread

Page 742 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Abwx

Lifer
Apr 2, 2011
10,847
3,297
136
And I do think they have a good chance of hitting 30% in 18 months. I don't see it ending there either.

This quarter eventually, the last numbers from AMD and Intel put the former at 27.7% revenue marketshare, and it would require 18 months to get to 30%..?...

Their sales amount to 38.45% of Intel s ones, that is 1.655 bn and 4.305 bn respectively.
 

Geddagod

Golden Member
Dec 28, 2021
1,147
1,003
106
This quarter eventually, the last numbers from AMD and Intel put the former at 27.7% revenue marketshare, and it would require 18 months to get to 30%..?...

Their sales amount to 38.45% of Intel s ones, that is 1.655 bn and 4.305 bn respectively.
The guy was referring to market share of chips sold , like the amount of chips, not revenue market share.
However most estimates of number of chips sold market share put AMD at 30% or higher by the end of 2023, at least from what I read. GNR should help but I don't think it's going to slow AMD too much.
 

Harry_Wild

Senior member
Dec 14, 2012
830
150
106
Will, getting antsy and may just buy the 13900 at MSRP of $549.99 but is now marked up to $599.99 that is in stock now. The 13600 is suppose to have MSRP of $255.00. The 13500 which is $232.99 MSRP can be found at $330.99 in stock!

Which should I buy or should just wait?😀
 
Last edited:

eek2121

Platinum Member
Aug 2, 2005
2,904
3,906
136
Arrow Lake-S uses TSMC 3nm according to Raichu and both Arrow and Lunar seems to use Lion Cove, Intel likely could use both TSMC 3nm or Intel node. Don't forget that Arrow Lake-P was planned to use TSMC 3nm as well, we know this from an early roadmap.

For the GPU, sure, but the cores themselves will stay on an Intel node.
 
Nov 8, 2022
43
77
51
I am not sure exactly what you mean, but I am saying that AMD has far superior server chips, and has for years, and this is finally supported by their ever increasing market share. I don't see that changing until Intel finally has a decent server ship. And I do think they have a good chance of hitting 30% in 18 months. I don't see it ending there either.

ok, i will explain, i mostly agree to the above, but i dont agree to the sentiment below
But who cares about 8 memory channels. When you are facing a CPU core thats almost as fast per core, takes less watts to run, AND has 50-100% more cores and 50% more memory channels, you have no chance.
this thread is about future rapids, and here we had a discussion about emerald vs saphire, yet you dismiss it with a "who cares?" "you have no chance!"

so i am saying that hundreds of data centers should care, why shouldn't they care? why is it irrelevant to complain about emerald having same memory channels as sapphire? do you recognize that more people will (most likely) buy emerald then genoa ?
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,434
136
ok, i will explain, i mostly agree to the above, but i dont agree to the sentiment below

this thread is about future rapids, and here we had a discussion about emerald vs saphire, yet you dismiss it with a "who cares?" "you have no chance!"

so i am saying that hundreds of data centers should care, why shouldn't they care? why is it irrelevant to complain about emerald having same memory channels as sapphire? do you recognize that more people will (most likely) buy emerald then genoa ?
I guess I am saying that regardless of which of the 2 choices has 8 or 12 channels of DDR4 or DDR5, the performance, and the performance/watt and the wattage make any current Intel server CPU a BAD choice. If data center managers don't get that, its their problem to deal with the power bills and buying more servers or CPU to deal with the capacity.
 
  • Like
Reactions: Grazick

Geddagod

Golden Member
Dec 28, 2021
1,147
1,003
106
That's not what all current indicators suggest. Questionable whether even the GPU will use N3.
If the iGPU was battlemage based, architecturally, then I doubt it's gonna be on N3. Because that would mean either the discrete version of battle mage was N3 or they backported the architecture to an older node for discrete cards. The discrete battle mage cards being on N3 doesn't really make sense in the 2024 timeframe for Intel (and idek if TSMC is ready for 3nm GPUs in 2024). And backporting also doesn't really make sense in light of recent Intel communication about streamlining and simplifying their GPU architecture between markets.
 

Exist50

Platinum Member
Aug 18, 2016
2,445
3,043
136
And backporting also doesn't really make sense in light of recent Intel communication about streamlining and simplifying their GPU architecture between markets.
I think that's the key bit. I can completely believe they intended to have a flagship big iGPU product on N3 at some point, but they've been slashing budgets for CCG and AXG severely, and that kind of low ROI halo product would be the first to be cut.
 
  • Like
Reactions: ftt and Geddagod

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
If the iGPU was battlemage based, architecturally, then I doubt it's gonna be on N3. Because that would mean either the discrete version of battle mage was N3 or they backported the architecture to an older node for discrete cards. The discrete battle mage cards being on N3 doesn't really make sense in the 2024 timeframe for Intel (and idek if TSMC is ready for 3nm GPUs in 2024). And backporting also doesn't really make sense in light of recent Intel communication about streamlining and simplifying their GPU architecture between markets.

TSMC should be ready for N3/N3e by next year. At least according to the last time I checked, which admittedly was awhile back.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
N3E is supposed to be ready by H2 this year.

Yeah thought so. The only reason why I could see Intel not wanting to use N3-family nodes by next year is that it may be a bit dated by then for a new, competitive dGPU product. But Intel has made significant financial commitments to take N3 wafers from TSMC, so hey smoke em if you got em right?
 

Exist50

Platinum Member
Aug 18, 2016
2,445
3,043
136
Yeah thought so. The only reason why I could see Intel not wanting to use N3-family nodes by next year is that it may be a bit dated by then for a new, competitive dGPU product. But Intel has made significant financial commitments to take N3 wafers from TSMC, so hey smoke em if you got em right?
What would be better than N3 next year? Doubt they're going to jump on 20A or even 18A for GPUs.
 
  • Like
Reactions: Lodix

Geddagod

Golden Member
Dec 28, 2021
1,147
1,003
106
Does anyone know what architecture Intel plans to use in their "low power" island in MTL SOC tile?
I just find it really interesting that an Intel architecture will be on TSMC 6nm. If it is a previously used architecture, it would make Intel vs TSMC node comparisons much, much more interesting.
Also, Intel 18A lion cove on ARL mobile vs TSMC 3nm for ARL desktop would also be really fun to see.
 

Exist50

Platinum Member
Aug 18, 2016
2,445
3,043
136
Does anyone know what architecture Intel plans to use in their "low power" island in MTL SOC tile?
I just find it really interesting that an Intel architecture will be on TSMC 6nm. If it is a previously used architecture, it would make Intel vs TSMC node comparisons much, much more interesting.
Also, Intel 18A lion cove on ARL mobile vs TSMC 3nm for ARL desktop would also be really fun to see.
At minimum Gracemont, most likely Crestmont. But there are other factors at play with the MTL SoC die that will render an apples to apples comparison impossible. Lion Cove + Skymont on N3 vs 20A will be much more informative in that regard.
 

Geddagod

Golden Member
Dec 28, 2021
1,147
1,003
106
At minimum Gracemont, most likely Crestmont. But there are other factors at play with the MTL SoC die that will render an apples to apples comparison impossible. Lion Cove + Skymont on N3 vs 20A will be much more informative in that regard.
Different type of libs on SOC tile than the HP ones Intel is using? Maybe built using HD cells for the cores in the SOC tile? If not, I don't see why at least density wouldn't be comparable between them.
 

Exist50

Platinum Member
Aug 18, 2016
2,445
3,043
136
Different type of libs on SOC tile than the HP ones Intel is using? Maybe built using HD cells for the cores in the SOC tile? If not, I don't see why at least density wouldn't be comparable between them.
Where the amateurs might note irregularities, someone with enough domain knowledge could reasonably deduce what Intel's done. That said, I think it's unlikely that we'll see any accurate coverage of the topic publicly.
 

jpiniero

Lifer
Oct 1, 2010
14,509
5,159
136

A comparison of Sapphire Rapids and Granite Rapids sockets.
 
  • Like
Reactions: Saylick

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
What would be better than N3 next year? Doubt they're going to jump on 20A or even 18A for GPUs.

Maybe whatever comes after 3GAP? But I might be giving Samsung too much credit, in multitudinous ways. I had thought N2 might be ready by late 2024, but no, that's 2025. 2024 may be rather unexciting node-wise.
 

Doug S

Platinum Member
Feb 8, 2020
2,201
3,405
136
H2 at that. Good odds we don't see anything N2 other than Apple until 2026. We'll probably see mostly N3E in '24 and N3P in '25.

The first N2 barely shrinks at all so we'll have to see how much interest there is for it. It is quite possible there will be some sort of cost optimized N3 shrink (i.e. similar to N6/N4...N2.5?) denser than N2 that customers who care more about cost would prefer.

Until they add in the backside/buried power rails in the next iteration of N2 I don't see it as a mass market node. Apple is one of the few companies that can afford to be less concerned about cost since they don't have a low end to worry about, but even if they might not jump on board the first N2 if the availability isn't right for their product launches.
 

Exist50

Platinum Member
Aug 18, 2016
2,445
3,043
136
The first N2 barely shrinks at all so we'll have to see how much interest there is for it. It is quite possible there will be some sort of cost optimized N3 shrink (i.e. similar to N6/N4...N2.5?) denser than N2 that customers who care more about cost would prefer.

Until they add in the backside/buried power rails in the next iteration of N2 I don't see it as a mass market node. Apple is one of the few companies that can afford to be less concerned about cost since they don't have a low end to worry about, but even if they might not jump on board the first N2 if the availability isn't right for their product launches.
N2 might not provide much density, but the PnP numbers are quite compelling. I'm sure there'll be plenty of demand for it in 2026.
 

jpiniero

Lifer
Oct 1, 2010
14,509
5,159
136

WTFTech claims the embargo on Sapphire Rapids-W ends on the 15th and reviews on the 22nd.