Discussion Intel Nova Lake in H2-2026: Discussion Threads

[Khato]

How about NVL pricing?
That seems a lot more difficult to predict for any other Nostradami.

Indeed, especially since there will be the new tiers of large LLC skus. Will they just be 1.2x the price for 1.2x the gaming performance?

 

[reaperrr3]

I absolutely admire your optimism wrt pricing.

...Fair point
At least wrt launch prices.

How about NVL pricing?
That seems a lot more difficult to predict for any other Nostradami.

Yeah, because if the rumored 10-20% ST boost for the P cores is true, NVL is still unimpressive for gaming, unless they fixed any and all latency issues.

Zen6 will imo have at least a 25-35% gaming perf uplift, from ~7-12% higher core IPC (before L3), a 5-15% additional "IPC" boost in gaming from the 50% bigger L3, and at least 10% higher clocks.

And that's before considering that CCD vs. NVL compute chiplet, AMD is increasing core count while Intel is not (I still consider the NVLx2 models a desperation measure to at least stay ahead in MT).

Unless NVL is a bigger improvement than expected, I have a hard time seeing any of the mainstream parts (even regular 8+16 without extra L3) selling above ~300$.
And DIY people may still not care regardless of price, if Zen6 wins in gaming.

 

[511]

Zen6 will imo have at least a 25-35% gaming perf uplift, from ~7-12% higher core IPC (before L3), a 5-15% additional "IPC" boost in gaming from the 50% bigger L3, and at least 10% higher clocks.

This is even more of a cope I doubt Zen 6 will match Zen5X3D same for vanilla NVL my guess
NVL ~ Z6 Vanilla in gaming
NVL bLLC will be slightly slower than Zen6X3D
For

Unless NVL is a bigger improvement than expected, I have a hard time seeing any of the mainstream parts (even regular 8+16 without extra L3) selling above ~300$.

Yeah not happening $400-450 for 8+16 at the least more expensive node btw both for Zen6/NVL.

 

[adroc_thurston]

...Fair point
At least wrt launch prices

It'll be better value but I'd expect like $479 or 500 buck for a 12c.

 

[SteinFG]

Indeed, especially since there will be the new tiers of large LLC skus. Will they just be 1.2x the price for 1.2x the gaming performance?

Nah, most likely flat +$150 like AMD does right now. +$200 if they go with 2X big tiles instead of 1 big + 1 regular

 

[OneEng2]

Let's not pretend 18A and friends are magically super-cheap (they aren't).

Agree.

Even though 18A ain't cheap but they are eating double margin on 18A products unlike N2.

The bigger issue is the TCO for Intel for 18A. Yes, individual wafers are likely more expensive for Intel on 18A than TSMC on N2 (longer, more complex process involved for BSPDN IIRC)... but I think the care and feeding of new nodes is a much larger cost.

Intel and AMD disagrees with you since they are releasing ~50T CPUs for the consumer DT market in 2026.

Cope with it and embrace the future.

AMD once tried to market "more cores" while Intel had the low/single core advantage. Ironic how little things have changed other than the companies position on which is more important .

You do understand that top SKUs are maybe 2% of the overall lineup sales?
They exist for bragging rights.

DIY looks to be ~10%:

https://www.reddit.com/r/hardware/comments/1kocozp/q1_2025_market_share_for_x86_processors

This is much higher than I would have guessed. I would have been with you on the 2% ... or less.

If you only intend to do mid-range gaming, I agree you can stay on old peasant CPUs with 6-8C CPUs. You're better off spending your money on the GPU.

Gaming is mostly limited by the GPU. Still, it isn't really a matter of actual advantage, but perceived advantage .... and gamers perceive they want the fastest frame rates money can buy.

Intel is screwed big time here in gaming, IMO. Their P/E core architecture is going to kill them. If AMD goes to 12 core CCD, they can have 12 big cores on one CCD with v-cache. The ideal response: more cores with v-cache, and no need to allocate the threads to the proper cores.

Intel is "screwed" because they lost DC where the margins are astronomical to AMD ..... almost over night and haven't done much in the new generation architecture to fix the problem.

Admittedly, HEDT/Gaming does command decent margins as well, and Intel is suffering here as you indicate.

I tried my best not to talk about the competition. But it's true, Intel probably didn't envision the 265K as midrange. It simply ended up there. And it's not for lack of MT performance. It seems like some other factor might be causing Intel to price these parts so low (it can't game).

I think Intel had to release ARL and move on. I believe that they may surprise some folks when they get to a revision that alleviates the latency problems.... at least I sure hope so. We need 2 good and healthy companies in this market!

My prediction
Ryzen 7 - 12C, 48MB L3, N3P: ~$399
Ryzen 5 - 10C, ?, N3P: ~$299
Ryzen 3 - 8C, 32MB L3, N4P: ~$199

I will bookmark the page and see how correct my prediction is, just like I bookmark adroc said we won't get more than 16-core in the future....

Possibly! Lets look at it next year .

 

[511]

The bigger issue is the TCO for Intel for 18A. Yes, individual wafers are likely more expensive for Intel on 18A than TSMC on N2 (longer, more complex process involved for BSPDN IIRC)... but I think the care and feeding of new nodes is a much larger cost.

It's Fine tbh they will milk it like 14nm/Intel 7 even if IFS stops pursuing External Customer and disbands 14A

 

[Thunder 57]

My prediction
Ryzen 7 - 12C, 48MB L3, N3P: ~$399
Ryzen 5 - 10C, ?, N3P: ~$299
Ryzen 3 - 8C, 32MB L3, N4P: ~$199

I will bookmark the page and see how correct my prediction is, just like I bookmark adroc said we won't get more than 16-core in the future....

$200 is too low. Bump all those by $50 would be my guess. Maybe even more (hopefully not).

 

[alcoholbob]

As it is now with Nova Lake Intel has to seriously fix the cache latency of their chiplet design. It's not like their monolithic design was that far behind. If you gave Raptor Lake a big cache arguably it would beat the 9800X3D right now as they are already faster in gaming than the non-X3D Zen 5 CPUs, so the question really remains is, can Intel actually make a chiplet architecture that doesn't suck worse than 12th gen? If they can figure it out, maybe they can be competitive with Zen 6 X3D. If it's just Arrow Lake style dog poo dressed up with BLLC, then they are just gonna get burned again.

Of course since this is MLID everything has to be taken with a grain of salt, but he had reported in late 2024 that Arrow Lake originally had a BLLC-style halo part with Adamantine L4 that was cancelled by Pat at the last minute, because supposedly the performance was disappointing. If that's the case and Intel's big cache parts don't have the same boost as X3D, then they probably are doomed.

 

[511]

It's MLID he has been saying Adamantine with B770/MTL and what not 40% for ARL so it's better to ignore him

 

[DavidC1]

DIY looks to be ~10%:

https://www.reddit.com/r/hardware/comments/1kocozp/q1_2025_market_share_for_x86_processors

Why would DIY be only be 2%? It's not just bleeding edge enthusiasts building their own computers. Plus it's a high margin/revenue market per unit sold. There's very little reason to market it for other segments, as they are vanilla and boring - no passion, which is what drives real changes.

Gaming chairs, monitors, mouse, keyboard, they would not exist if it was a mere 10%. This was an obvious conclusion.

Also:

This also results in that the average selling price of desktop processors in the overall market must be +73% higher than for mobile processors.

This suggests the actual share of DIY in desktop market is higher than 10%, I'd say it's closer to 20-25%.

Passion, and enthusiasm is what drives people. That's what DIY market has. Kill that and the whole market, not just CPUs will tank hard. Also, DIY will be more resilient in harder times because it's cheaper for the user unlike a laptop, which is a throwaway for every part.

3D Center also says this:

But that development also shows how difficult it is for AMD to steal substantial market share from Intel – even in times when Intel is not doing well in business.

Right, because right now Intel is still holding on. There's a tipping point though, when things escalate at a dramatically increased pace. If you look at Intel's recent revenue figures, it's pretty much the client market holding them up. So whatever little they are losing on client will hit them hard from now on. They made lot of money on server, but not anymore. Easy to gain, easy to lose.

 

[LightningZ71]

AMD is, and continues to be, deliberately volume limited at retail. They would rather build fewer units at a higher ASP than to build an excess of units at a lower ASP. For AMD, this isn't a bad thing, as it keeps their ASPs high and their wafer buys under control. It sucks for us though, as it keeps our costs higher.

So long as this is their strategy, Intel will retain a big share at retail/DIY as there is more market than what AMD is willing to produce for. So long as Intel retains a large share, OS vendors are going to continue to target their optimization efforts at Intel. So long as they retain their large share, laptop OEMs are going to give AMD piss poor efforts on their products. Those things are an animal of AMD'S making.

 

[jpiniero]

If anything, Intel's DIY share right now is kinda crap. Which shouldn't be a surprise given how much better Zen XD is in games.

 

[Thibsie]

If anything, Intel's DIY share right now is kinda crap. Which shouldn't be a surprise given how much better Zen XD is in games.

I don't think this is the only reason.

 

[511]

I don't think this is the only reason.

Poor Platform longetivity and Some Techtubers promoting 7600 vs 14600K just cause 14600K has no upgrade path but 14600K has BF6 as well and both are similar in price

 

[OneEng2]

As it is now with Nova Lake Intel has to seriously fix the cache latency of their chiplet design. It's not like their monolithic design was that far behind. If you gave Raptor Lake a big cache arguably it would beat the 9800X3D right now as they are already faster in gaming than the non-X3D Zen 5 CPUs, so the question really remains is, can Intel actually make a chiplet architecture that doesn't suck worse than 12th gen?

I mostly agree. I think that Intel's bigger issue is the revenue and profit loss in DC where they are getting positively hammered.

It is also an issue that Intel is bested in the high performance, high profit desktop and DIY.

AMD is, and continues to be, deliberately volume limited at retail. They would rather build fewer units at a higher ASP than to build an excess of units at a lower ASP. For AMD, this isn't a bad thing, as it keeps their ASPs high and their wafer buys under control. It sucks for us though, as it keeps our costs higher.

So long as this is their strategy, Intel will retain a big share at retail/DIY as there is more market than what AMD is willing to produce for. So long as Intel retains a large share, OS vendors are going to continue to target their optimization efforts at Intel. So long as they retain their large share, laptop OEMs are going to give AMD piss poor efforts on their products. Those things are an animal of AMD'S making.

I completely agree. It's looking like AMD's strategy is to capture the margins at the high end with state of the art N2 process chips, and to let later generation chips on later generation processes address down market.

My only concern with this is .... well .... what does AMD do with that 12c CCD that only yields out 6c? It isn't like it is worthless, but how to position the price against last years 8c?

 

[jpiniero]

My only concern with this is .... well .... what does AMD do with that 12c CCD that only yields out 6c? It isn't like it is worthless, but how to position the price against last years 8c?

OEM only fodder.

 

[LightningZ71]

I mostly agree. I think that Intel's bigger issue is the revenue and profit loss in DC where they are getting positively hammered.

It is also an issue that Intel is bested in the high performance, high profit desktop and DIY.

I completely agree. It's looking like AMD's strategy is to capture the margins at the high end with state of the art N2 process chips, and to let later generation chips on later generation processes address down market.

My only concern with this is .... well .... what does AMD do with that 12c CCD that only yields out 6c? It isn't like it is worthless, but how to position the price against last years 8c?

There are a few options:
1) it can be used in lower end EPYC SKUs.
2) it can be used with Medusa Premium as the extra CCD, which gives 6+4+4+2, which is still a heck of a lot on a laptop.
3) depending on the cache enablement, it could be a big cache per core 6+6 4xxx series.
4) garbage tier Ryzen 3 part. 6 cores, 24 MB L3, maybe no SMT either. Maybe combined with crippled IOD, single channel ram, only PCIe 4.0, half USB...

 

[OneEng2]

There are a few options:
1) it can be used in lower end EPYC SKUs.
2) it can be used with Medusa Premium as the extra CCD, which gives 6+4+4+2, which is still a heck of a lot on a laptop.
3) depending on the cache enablement, it could be a big cache per core 6+6 4xxx series.
4) garbage tier Ryzen 3 part. 6 cores, 24 MB L3, maybe no SMT either. Maybe combined with crippled IOD, single channel ram, only PCIe 4.0, half USB...

Just a clarification .....

Do we know that all 12c CCD's are created equal? Can you take the 12c CCD intended for a desktop and put it into a laptop? A server?

 

[LightningZ71]

Used to be a certainty that they were the same. Now, we know that some CCDs are made with a particular market in mind. Though, 6 core EPYC CCDs can just be used in lower end products, and the same with desktop ones.

 

[adroc_thurston]

Can you take the 12c CCD intended for a desktop and put it into a laptop?

Yes.

A server

No.

 

[igor_kavinski]

No.

They sell servers for 4 core 8 thread Cascade Lake. My idiot IT guy bought one Dell for over $2000. I was literally dismayed to see it having only 4 cores. I asked him and his answer was, it will run only one application. I honestly want to go around the world and make sure idiots like him never have an IT job again.

 

[adroc_thurston]

They sell servers for 4 core 8 thread Cascade Lake. My idiot IT guy bought one Dell for over $2000. I was literally dismayed to see it having only 4 cores. I asked him and his answer was, it will run only one application. I honestly want to go around the world and make sure idiots like him never have an IT job again.

Infinite reading comprehension levels.
(my point is that DT CCD != Server CCD. They're different things).

 

[igor_kavinski]

(my point is that DT CCD != Server CCD. They're different things).

Elaborate please. At higher core counts, yes it does matter. But for something like 16 to 32 cores, why not have desktop CCDs and save the cost? It's just the IOD that needs to be server class and support more RAM channels.

 

[adroc_thurston]

why not have desktop CCDs and save the cost?

Because you still need a different TO even for the same design for server.
Server CCDs target different v/f ranges, with different Vts and considerably different binning targets.