Discussion Intel Nova Lake in H2-2026: Discussion Threads

[inquiss]

man this kid really expect everyone to spoonfeed him.

money.

Mainstream will move from 6/8 to 8/10/12 and that will be it. You know that, I know that, but some people, they don't know that.

 

[branch_suggestion]

Oh, you've got access to the Intel/AMD price list for NVL-S and Zen6? Link to source? Oh, and by the way, you forgot to provide the link to the source for your previous 2% market share claim, so you can add that as well. Unless it was all just guessing/BS as usual.

Also, if you really think the 16C SKUs will not be provided at mainstream price, then what do you think the 52C CPUs will cost? Remember that 52C will be on the consumer DT platform, thus for consumers. So for it to make sense, it must be priced accordingly.

The price of the 12c X3D will dictate the market. I'm betting on $599.

 

[Thunder 57]

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 OTOH, will be stuck with 8 P cores per chiplet, and will have to either accept 2 chiplets to get more P cores, or assign the gaming to E cores, (or a mixture of both). This seems a horrible solution for a company that is already getting trounced in gaming, partly because of too much latency in their tile structure. Maybe, if the new v-cache equivalent actually materializes, it can compensate for this, but TBH, it seems like a scheduling nightmare with loads of extra latency compared to AMD with 12 homogenous cores on one CCD.

What is stopping Intel from putting 12 P cores on a tile? Size? Cost? Lack of vision?

 

[desrever]

What is stopping Intel from putting 12 P cores on a tile? Size? Cost? Lack of vision?

why bother?

 

[adroc_thurston]

What is stopping Intel from putting 12 P cores on a tile? Size? Cost? Lack of vision?

Their big cores just aren't competitive in PPA, and Atoms are very okay in cinememe@mm^2.

 

[gdansk]

Sweet. Now we'll soon get 16C at nice mainstream price too

It's already been done. The 265K/KF is 20 cores and has been firmly mainstream at $260 to as low as $210.

But you miss the goal of dual tile parts. It gives Intel more room to play SKU games to push higher margin parts. Much like dual CCD gave AMD a great deal of room to push stupid parts without decreasing the prices of 8 core parts significantly. If my memory is correct AMD actually used this approach to increase the price of 8 cores parts from Zen 1 to Zen 5.

I do not expect midrange ~$260 Nova Lake to offer more than 20% more Cinememe R23 per $$ than the 265K. Which is a normal generational uplift...

 

[adroc_thurston]

The 265K/KF is 20 cores and has been firmly mainstream at $260 to as low as $210.

That's actually not sustainable pricing for that SKU but it's not like Intel has a choice.

 

[gdansk]

That's actually not sustainable pricing for that SKU but it's not like Intel has a choice.

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).

 

[Saylick]

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 OTOH, will be stuck with 8 P cores per chiplet, and will have to either accept 2 chiplets to get more P cores, or assign the gaming to E cores, (or a mixture of both). This seems a horrible solution for a company that is already getting trounced in gaming, partly because of too much latency in their tile structure. Maybe, if the new v-cache equivalent actually materializes, it can compensate for this, but TBH, it seems like a scheduling nightmare with loads of extra latency compared to AMD with 12 homogenous cores on one CCD.

The other thing is that with 12 Zen 6 cores + 64MB of Infinity $, "legacy" games that still only leverage 8 threads continue to benefit from the shared L3 from the additional 4 cores. This benefit stacks on top of whatever ST improvements Zen 6 brings to the table.

 

[adroc_thurston]

The other thing is that with 12 Zen 6 cores + 64MB of Infinity $, "legacy" games that still only leverage 8 threads continue to benefit from the shared L3 from the additional 4 cores. This benefit stacks on top of whatever ST improvements Zen 6 brings to the table.

a) 48M L3
b) the V$ pile will also be a bit larger

 

[Saylick]

a) 48M L3
b) the V$ pile will also be a bit larger

a) Yes, that's my point. If a game still only uses 8 cores, a 12 core CCD with it's 48 MB L3 will have an advantage over an 8 core CCD with only 32 MB L3.
b) Oh, good, so a new V$ die then to go with Zen 6. Hopefully, it will provide a tripling of L3 like it did for Zen 3, 4, and 5.

 

[adroc_thurston]

If a game still only uses 8 cores, a 12 core CCD with it's 48 MB L3 will have an advantage over an 8 core CCD with only 32 MB L3.

eh, incremental.

Oh, good, so a new V$ die then to go with Zen 6. Hopefully, it will provide a tripling of L3 like it did for Zen 3, 4, and 5.

It's always a new V$ die.
the TSV layout is always different.

 

[CakeMonster]

a) Yes, that's my point. If a game still only uses 8 cores, a 12 core CCD with it's 48 MB L3 will have an advantage over an 8 core CCD with only 32 MB L3.

This was widely discussed with Z3, people speculated that the 5900X would have a theoretical advantage in some scenarios over the 5950X. It was tested (I can't remember where) and nobody could find any scenario where it benefited.

 

[Khato]

Back on the subject of NVL, why not guess how much improvement the large-LLC variant is going to bring on the gaming front compared to ARL? 12%? 24%? 36%? 48%?

 

[adroc_thurston]

Back on the subject of NVL, why not guess how much improvement the large-LLC variant is going to bring on the gaming front compared to ARL? 12%? 24%? 36%? 48%?

no one knows because no one knows how bad NVL L3$ perf is.

 

[gdansk]

Back on the subject of NVL, why not guess how much improvement the large-LLC variant is going to bring on the gaming front compared to ARL? 12%? 24%? 36%? 48%?

How much do they need?

 

[511]

That sure as hell happened with 16c parts over the span of 6 years.

14C 14600K/16C 12900K 😛

 

[511]

How much do they need?

Considering 20% faster for 9800X3D they need 40%

 

[adroc_thurston]

14C 14600K/16C 12900K 😛

That's not a real 16c.

 

[511]

That's not a real 16c.

Those are real cores present in silicon get your facts checked 😛

 

[gdansk]

Those are real cores present in silicon get your facts checked 😛

If only someone had been wise enough to sell their earlier parts as 4P+4E they wouldn't have had to settle that lawsuit. 🤔
Unfortunately all those shared FPU cores performed poorly so such a fiction wasn't possible.

 

[reaperrr3]

eh, incremental.

Yes, but thanks to this the effective gaming perf uplift of non-X3Ds could exceed Zen4's over Zen3, and that's enough of a deal that I can absolutely see folks that can't afford (or are too stingy for) X3D SKUs finally upgrading with this, and the Zen6 equivalents of 9600X/9700X being much more successful than their Zen5 predecessors.

Like
$399 - 10800X - 12C, 48 MB L3, 6.3 Ghz turbo, ~10-20% higher gaming IPC
$349 - 10700X - 10C, 48 MB L3, 6.2 Ghz turbo, ~10-20% higher gaming IPC
$279 - 10600X - 8C, 48MB L3, 6.1 Ghz turbo, ~10-20% higher gaming IPC

might actually sell at those prices, unlike anything Zen5 below the 9800X3D.
I mean, such a 10600X would get people near-9800X3D gaming perf at ~9600X launch price.

To be a little on topic: If NVL doesn't bring a massive gaming perf boost, Intel might be (even more) toast in mainstream desktop, on top of already being toast in high-end.

This was widely discussed with Z3, people speculated that the 5900X would have a theoretical advantage in some scenarios over the 5950X. It was tested (I can't remember where) and nobody could find any scenario where it benefited.

Multi-threading in games is usually inherently non-uniform, it's more like the core gameplay loop is on one heavy main thread, then secondary stuff like UI gets a separate thread, cosmetic non-critical stuff like daily NPC routines gets its own thread(s), audio get its own 1-4 threads, then texture- or world-streaming gets its own 2-8 threads etc.
As a result, quite often it's only that 1 heavy main thread that greedily takes any L3 it can get, and since both 59XX have the same amount of L3 per CCD, there's no difference because the maximum amount of L3 accessible by 1 thread is the same.

Zen6 vs. Zen5 will be quite different, because the maximum amount of L3 that can be hogged by such a heavy main thread goes up by 50%.

Of course, I can't rule out that AMD will disable 1/3 of L3 on octa-cores to upsell 10C/12C SKUs.

 

[adroc_thurston]

$399 - 10800X - 12C, 48 MB L3, 6.3 Ghz turbo, ~10-20% higher gaming IPC
$349 - 10700X - 10C, 48 MB L3, 6.2 Ghz turbo, ~10-20% higher gaming IPC
$279 - 10600X - 8C, 48MB L3, 6.1 Ghz turbo, ~10-20% higher gaming IPC

I absolutely admire your optimism wrt pricing.

 

[Tigerick]

Yes, but thanks to this the effective gaming perf uplift of non-X3Ds could exceed Zen4's over Zen3, and that's enough of a deal that I can absolutely see folks that can't afford (or are too stingy for) X3D SKUs finally upgrading with this, and the Zen6 equivalents of 9600X/9700X being much more successful than their Zen5 predecessors.

Like
$399 - 10800X - 12C, 48 MB L3, 6.3 Ghz turbo, ~10-20% higher gaming IPC
$349 - 10700X - 10C, 48 MB L3, 6.2 Ghz turbo, ~10-20% higher gaming IPC
$279 - 10600X - 8C, 48MB L3, 6.1 Ghz turbo, ~10-20% higher gaming IPC

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....

 

[gdansk]

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