Friday, July 3rd 2026
Intel Readies Two 22-Core "Nova Lake-S" SKUs with 108 MB bLLC Cache for Gaming
According to well-known Intel leaker Jaykihn, Intel is preparing two "Nova Lake-S" CPU models with a 22-core configuration, enhanced by the big Last Level Cache (bLLC). These processors are part of the Core Ultra 5 400S family, and Intel is developing two mid-range configurations, each with 22 cores. Both models will feature 6 "Coyote Cove" P-Cores, 12 "Arctic Wolf" E-Cores, and 4 LPE-Cores. This design includes a single compute die aimed at mid-range gaming rigs. Alongside the 22 cores, Intel plans to add an extra 108 MB of bLLC cache to boost gaming performance, similar to AMD's approach with its Ryzen X3D gaming CPUs, which currently lead in gaming performance. Intel aims to match raw CPU compute power with newer CPU cores and use bLLC to improve latency-sensitive tasks like gaming. The only difference between the two new models is that one is an unlocked 125 W TDP version, part of the "K" overclockable family, while the other is a 65 W base TDP model.
Recently, it has been observed that Intel's upcoming Z9x0 platform for "Nova Lake-S" will support up to 52 cores in a single LGA-1954 socket. Some early rumors suggest a PL2 power draw of 474 W, which is a significant increase over previous designs. This is understandable, considering that the design brings workstation-like core counts into the consumer sector for the first time. Regarding the socket, Intel's LGA-1954 has been pictured with a 2-lever independent loading mechanism (ILM), similar to what Intel's HEDT sector uses. This is typically reserved for larger CPU packages, and with the pin count now nearing 2,000 socket pins, it is a necessary choice for Intel. Importantly, the first "Nova Lake-S" CPU SKUs are scheduled for release in just a few months, likely leading to more rumors as we approach the final launch.
Source:
Jaykihn on X
Recently, it has been observed that Intel's upcoming Z9x0 platform for "Nova Lake-S" will support up to 52 cores in a single LGA-1954 socket. Some early rumors suggest a PL2 power draw of 474 W, which is a significant increase over previous designs. This is understandable, considering that the design brings workstation-like core counts into the consumer sector for the first time. Regarding the socket, Intel's LGA-1954 has been pictured with a 2-lever independent loading mechanism (ILM), similar to what Intel's HEDT sector uses. This is typically reserved for larger CPU packages, and with the pin count now nearing 2,000 socket pins, it is a necessary choice for Intel. Importantly, the first "Nova Lake-S" CPU SKUs are scheduled for release in just a few months, likely leading to more rumors as we approach the final launch.

279 Comments on Intel Readies Two 22-Core "Nova Lake-S" SKUs with 108 MB bLLC Cache for Gaming
However I do want to get as much for my money as I can which is why I value platform longevity, I'd get more excited for Intel CPU's if they'd keep a socket around longer as a pro-consumer move instead of changing sockets to satisfy OEM's and the motherboard makers. Due to the cost of RAM I would much rather keep my kit of RAM and motherboard to have the option for a drop in CPU upgrade.
What I don't understand is why people are calling 8 cores weak, those "weak" 8 cores are outperforming CPU's with double the cores in games. Prices aren't extreme either unless you're buying the 9850X3D or one of the higher core count high end CPU's though I consider those to be more of a workstation CPU. I don't quite understand the desire for a CPU with a bunch of weaker e-cores either, especially for a high end enthusiast system.
Furthermore, the e core IPC in novalake is similar to zen6, with the highest ppa, so intel can provide more MT than amd, making the overall design very high-end.
If it's a price argument (which you didn't specify, I just assume), then you have a point. There's no reason to overspend on the likes of the 9800X3D. The 7700 non-X or 9700X are just fine as far as 8 cores go.
The money spent is irrelevant since these are all relatively expensive high end cpu's.The used market is always there for those wanting a price/performance deal.
I used to always sell my old hardware to buy the latest cpu's, it used to be worth it but it's been increasingly not worth the hassle.So because the CPU is cheap, somehow it's a better value to buy a new CPU and board every 2 years?
I can understand that some may need more MT performance, but I don't get the logic of buying a whole new system every 2 years because the CPU is slightly cheaper.Enthusiasts here are not the majority, the majority don't care about how many cores they have or saving a few seconds on an Excel sheet.I don't need to replace the CPU frequently, and no I don't understand what you're getting at.
With AM5 you could've bought a motherboard several years ago without the need to buy another one until Zen 7. Imagine if Intel kept LGA1700 around that long, however Intel users seem to decry having platform longevity is a bad thing because having the "latest features" is better even though the usefulness of a majority of those features is debatable.Those 8 cores aren't weak, it baffles me why people are saying this. www.techpowerup.com/review/amd-ryzen-7-9850x3d/19.html
Motherboards aren't cheap anymore, not sure where the logic is on this, any decent motherboard capable of overclocking is usually as expensive as a 270K+.
But we're moving the goalposts to OC performance? I though efficiency mattered, OCing throws the whole efficiency claims out the window, to each their own though. Not everyone only cares about MT though, for a gaming system a majority of people don't which is why X3D CPU's have been the best selling from several retailers.In 2022 LGA1700 was expensive too for a CPU that could compete with AM5, along with the added cost of getting a high end cooler and board that could OC a 13th gen CPU. Also buying a new CPU and board seems wasteful to me, rather than having the option to install a new CPU if I wanted to, not a requirement but a nice option to have.
Zen 6 may require more power, though the X3D variants have been efficient since AM4. Rumors of Nova Lake are saying it may consume more power also, which will drive up motherboard cost.
The loss of new features like PCI-e 6.0 or the latest thunderbolt revision won't matter to most people with a gaming PC. Not having PCI-e 5.0 isn't a significant loss either unless you need tons of fast storage, and the cost of filling every M.2 slot with a gen 5 SSD is too expensive now.
So it seems Intel users care about about buying and selling hardware more than any kind of hardware longevity, rather than apparently keeping the same CPU for 3-5 years. Motherboard resale value isn't of much importance to me, a used CPU is easier to sell than a motherboard, also used motherboards don't have a lot of resale value unless you're buying a limited edition high end board anyway.4 cores were weak when Intel stagnated with near a monopoly on the market, any high performance CPU isn't weak today. most things are a few seconds compared to something with nearly double the cores.
I'm not a youtube influencer or counting on loading an office program a few milliseconds quicker, so the tradeoff for faster more efficient gaming load for slightly less productivity performance is a worthy tradeoff IMO.
www.techpowerup.com/review/amd-ryzen-7-9850x3d/12.htmlOutside of a few outlying games, games are going to prioritize P cores. I prefer a CPU with all full fat cores, no chance of a game or application trying to use slower e-cores.
Do you know Nova Lake e-cores will perform the same as Zen 6? So far all we have are rumors, and that seems like a very optimistic rumor. If true then I doubt Intel is going to sell CPU's at a loss like Arrow Lake and ARL refresh.
It is just a motherboard. The reason people sell motherboards and cpus to upgrade to the next gen is the same as the reason you sell a cpu alone. and less money is lost when selling. If you look at the prices of intel cpus, intel's motherboard sockets make the cpus hold their value better.
And you can always use a cpu for a longer time instead of replacing it every 2 years. Since this is the case, we seem to agree with each other on this point. Then the vrm of zen4 am5 motherboards is barely enough to support zen6 unless you buy a high-end one. But when zen6 is released, mid-to-low-end motherboards will support zen6 very well. If you are thinking about upgrading at that time, you must spend a lot on the motherboard from the very beginning, which seems very uneconomical.For arrowlake, the main improvement comes from the uncore part. You can also undervolt and overclock, which is a common oc method for ryzen as well.
However ddr5 is expensive now, so using high-end Hynix memory for a new build and oc is not very realistic.You can actually calculate the vrm required for zen6. It will be higher than the 9900x and roughly requires a vrm that can satisfy the 9950x. This means most b650 cannot do it. For a mid-tier motherboard at the time of b650, you could only buy a low-end zen6 8 core, if that even counts as an upgrade.
The x3d actually requires a better vrm, even though the cpu itself is more efficient.
If you check the prices of old cpus like the 10900k and 11700k, you will know that intel cpus can still be sold for quite a bit after being used for a long time.
The motherboard doesn't matter at all. I saw many people throw away their old am4 motherboards to buy new am4 ones when zen3 was released because the old am4 motherboards were too poor. This situation still applies to b650.That is your personal opinion, but actual tests show that the hybrid approach can improve gaming performance by adding threads outside the 8 p cores. The zen6 12 core gaming performance will be better than the 8 core, which is the same way like intel hybrid works. Only a few games do not work well on hybrid, and intel has basically solved this problem through apo. As you can see, the apo list is very small, which shows that there are actually very few games with problems.
Today you can get 16 or 24 core CPUs for normal consumer PCs. 8 core CPU today is so weak as 1 or 2 core CPUs used to be years ago. I cannot imagine anybody being excited about 1 core CPU 10 years ago, even if it could run games well.
These are high performance CPU's being sold mostly to the gaming market, so unsurprisingly gaming performance comes first, a majority of enthusiasts aren't measuring Excel or the Adobe suite as a benchmark.
AMD is milking gamers VERY HARD - they have their udders all bloody and shrunken, yet they seem happy anyways.
in fact it is the next generation of arctic wolf/golden eagle.
, and is still divided into p/e core hybrid.
Hybrid is a common approach for all Silicon Valley hardware manufacturers, apple, amd, intel, qualcomm, which shows that hybrid has great advantages.
In fact intel provides 2x MT. after amd introduces 12 core, even amd fans will go 12 core, 8 core era will be finished
One study says:
Intel E-cores are simply reduced P-cores.
Schedulers almost always choose P-cores,
Intel Thread Director often makes wrong decisions, often makes mistakes(rarely classifies workloads as "vector heavy", rarely detects spin loops, inaccurately predicts IPC)
The result: E-cores are almost never the optimal choice.E-core is not the same as Zen 6, because Zen 6 physically has more internal resources, a wider pipeline, and a larger cache and SMT, which E core does not have. E core is a scaled down version of P core without SMT, while Zen-c is the same full Zen core with just reduced cache and frequency.
And "back then" this "a little slower" wasn't so noticeable, too;):Dnot "very", just a little slower.
I think Arcticwolf and Zen6 IPC are about the same. I believe I have reduced the IPC for E-core as much as possible, while increasing the IPC for Zen6 as much as possible.
This is because Raichu believes the 20% IPC growth for Arcticwolf is based on Darkmont, and Darkmont has 5% higher IPC than Skymont. Meanwhile, MLID believes Zen6 increases 6% IPC in SPEC FP, and now estimates it at 10% but not for any specific SPEC, which might be an AMD-style random IPC, and what I calculated is strict SPEC,so I increased it to 8% for it, which is an overestimation.
And I took the maximum value for the Zen5 and the average value for the Darkmont across 2 data sets.
This of course is not any serious data, just some simple calculations that can be done under the existing information.
---
I need to add one more point for people who support the following two points:
1 I do not want e core to disrupt core scheduling.
2 I do not need mt, or even st, I only need gaming performance
Because amd features smt, and smt undoubtedly disrupt core scheduling, and its poor performance reduces gaming performance, as tpu has relevant tests,
unlike e core which increases gaming performance
(even 13gen e core can increase gaming performance, let alone the latest e core architecture now which has made significant progress and is therefore better).
Therefore, the only way for you to be logically self-consistent is to turn off amd's smt, keep only physical threads, as this smt low-quality thread is disrupting your scheduling and reducing your gaming.
So did you do it, absolutely recommend you do it.
chipsandcheese.com/p/skymont-in-desktop-form-atom-unleashed
And later, skymont and all other ecores is not configured for 1-IPC workloads such as games
chipsandcheese.com/p/skymont-in-gaming-workloads
chipsandcheese.com/p/intels-lion-cove-p-core-and-gaming
chipsandcheese.com/p/running-gaming-workloads-through
The IPC of e core is significantly higher than zen4x3d, according to the picture, zen4 only has an advantage in 3 sub-items, while skymont leads in 18 sub-items.
Do you know what IPC is.
Write correctly or don't write.
There is no test sub-item called games in SPEC.
And e core can increase gaming performance by increasing the number of threads. Compared with 8 p core, 8p16e gaming performance is improved.
Some people keep claiming that e core affects gaming performance, are they just unwilling to face the real test results.
E cores are not ready for a "Unified Core" which is why NVL will still have P Cores.chipsandcheese.com/p/skymont-in-desktop-form-atom-unleashed
E cores are a way to get more MT performance out of a piece of silicon, that is all they are for.
For example 14600K (6P+8E) thank to E-cores has noticeably higher MT performance than 8 core AMD chips and thus are better universal PC processors.
This hybrid P + E cores architecture is not a nonsense, as some people try to portrait it, but it simply works well in what it tries to achieve - improving MT performance.
Zen6 (theoretical, because it is not yet released) is faster not only due to larger resources and higher frequencies, but also due to a better front-end, better branch prediction, better prefetching and a wider µop pipeline. This gives Zen6 both higher IPC and better real-world ST/MT performance. So, to claim that only Intel makes architectural quality improvements and AMD does not is again incorrect.
Disabling SMT does not make sense in the gaming context, because as I understand it you are relying on features of the Zen-c architecture, which AMD gaming processors do not have at all. Because Zen-c is not used in games, but only in EPYC servers: Bergamo (Zen4c), Siena (Zen4c mix) and Turin-Dense (Zen5c) and in some laptops.