• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Intel "Nova Lake‑S" Series: Seven SKUs, Up to 52 Cores and 150 W TDP

Ugh. I just want another 8p+0e SKU that runs at ~60C with a $35 cooler like my current 7800X3D. For the overwhelming vast majority of desktop PC users/gamers, 8P cores is the sweet spot, as most games still today don't use more than 1 or 2 cores anyway so after even 4 and especially 8 cores there are drastic diminishing returns for actual performance for real-world apps/games, these "52 cores" are great for rendering projects but complete overkill in the wrong direction for the vast majority of users Intel seems to be targeting.

I'd love to be wrong, maybe 2025 we'll finally get game engines and apps that can actually use all these cores but it seems unlikely.
Hard to say. There are already enough scheduling issues, and games are so latency sensitive that I don't know that they will ever want anything more than one type of core. Maybe they could leverage them for more NPC activity?
That RAM is unobtainium, and you won't find 10k MT/s any time soon at a reasonable price.
But even if we did, stretching so far would still only would reach 160 GB/s. It's a significant step up from 96 GB/s with DDR5-6000, but still much less than the 273 GB/s of the M4 Pro.
They need to either move to quad-channel or make a new DIMM format that is 128-bit.
Strix Halo goes up to 256GB/s, but really, that's largely for feeding the GPU. Back when the M1 Max launched, the CPU complex could pull as much as about 240GB/s from the unified memory, but that was using synthetic benches. Granted, if Intel does go to that many cores, then bandwidth is going to start mattering. I suspect the E-cores are going to get another performance uplift.
 
Those are your claims only. I'd much rather wait for actual reviews, than put my bets on some random guy's comments.
Oh absolutely. I'm just saying that based on what we saw with the first Core Ultra CPUs, lots of e cores absolutely performs better in multi-thread workloads than fewer fast HT P cores, so much so that Intel can match Zen 5 with fewer threads, just based on the efficient design of the e cores.
Lastly, please do blindly trust Intel's judgment of how many cores the average user needs, but don't forget it is the same company that gave you nothing but 4-cores on 7 generations of products.
Soooo... do you want consumers to have a choice on core count or not? I'm confused, because you're mad Intel has lots of cores on their premium i9 product, but you're also mad Intel didnt' have lots of cores on their premium products.
 
The 32 e cores will take space of eight P

Wait did I miss something? Skymont currently stacks about 3:1 E to P core ratio, with a full slice of 8 E-Cores consuming 1.34x more die space than 2 P-Cores. Has it been leaked/announced somewhere that the Arctic Wolf E-Core layout will be significantly smaller?

the M4 Max you're comparing to is a significantly more expensive platform than the upcoming Intel Nova Lake

Presumptuous, but may prove to be right. M4 Max Mac Studio can be had for about $1800 right now with current sales. Apple usually gets even more aggressive on refurb prices. Meanwhile I priced up an Ultra 9 285K+Z890+DDR5-8400 config for about $1000-$1100 without SSDs, video card, case, power supply, or OS.
 
Intel has gone insane with core counts. Can't imagine what these would be used for as very few tasks scale CPUs cores up like this. I feel like 8P + 8E + 4LPE is the sweet spot for any high end PC outside datacenter stuff. Oh well, here is to hoping they are more competitive to AMD next time. They reigned in their power consumption at least.
 
Wait did I miss something? Skymont currently stacks about 3:1 E to P core ratio, with a full slice of 8 E-Cores consuming 1.34x more die space than 2 P-Cores. Has it been leaked/announced somewhere that the Arctic Wolf E-Core layout will be significantly smaller?
The 12/13/14th gen e cores were a little smaller relatively, the next gen ones little bigger relatively, especially if you compare a full eight, but they are divided into stacks of four, interposed with P cores, instead of the old raptor Lake design of all the p then all the e. Regardless, 1:4 or 1:3.5, and whatever the ratio is for nova lake, the point I'm making is unchanged and it's napkin math. No need to get pedantic over exact 2:7 or 2:8 ratios, unless that data changes the underlying point, which it doesn't. It's better to have 8/16 P Cores and 16/32 E cores than 3-4/7-8 extra P cores but no E cores. This has been discussed to death, and the general consensus is anything that scales to nT does better with e cores, and the ST/latency sensitive stuff doesn't (at least on consumer platforms) really scale past eight. Even AMD realises this, after ARM and then Intel, with their Zen C cores.
Presumptuous, but may prove to be right. M4 Max Mac Studio can be had for about $1800 right now with current sales. Apple usually gets even more aggressive on refurb prices. Meanwhile I priced up an Ultra 9 285K+Z890+DDR5-8400 config for about $1000-$1100 without SSDs, video card, case, power supply, or OS.
All discussion on unreleased products is, by definition, presumptuous. We're operating under the presumption that the leaks are accurate and discussing them in the relevant thread. Unless you're implying that Intel's top consumer CPUs will go from $5-700 for decades, to significantly different pricing, which would be suicide in face of competition. You're also comparing against the base RAM M4 Max, which makes little sense to buy considering the types of task these chips excel at. Apple math means comparable RAM amounts add up to a lot more than "$1800" to get the 96/128 GB you would pair a workstation based on x86 consumer silicon with. Unless that $1800 was also napkin math, because a quick check reveals "starting at $1999."

1000018438.png


Just checked. 36 GB unified (VRAM + RAM) for $2k, $4k for 96 GB.
 
"Got any moar of dem Cores ?"

If so, I'll take 3,726,589.241 please, like, yesterday, hahahaha... :eek:..:roll:..:respect:
 
because you're mad Intel has lots of cores on their premium i9 product, but you're also mad Intel didnt' have lots of cores on their premium products.
First you claim to know what Intel's e-core bumped CPU's performance is going to be. Now you also claim I'm mad?

Are you basing those claims on what? Your crystal ball or the voices in your head? Come back when you have some facts to back you up.

Until then, customers indeed have a choice on core count, thanks to competition, not a stagnant architecture, or having to throw your motherboard in the trash every 1 or 2 generations.
 
Last edited:
First you claim to know what Intel's e-core bumped CPU's performance are going to be. Now you also claim I'm mad?

Are you basing those claims on what? Your crystal ball or the voices in your head? Come back when you have some facts to back you up.

Until then, customers indeed have a choice on core count, thanks to competition, not a stagnant architecture, or having to throw your motherboard in the trash every 1 or 2 generations.
While annoying for people that build their own PCs, how many normal people are impacted by this? Even when sockets have lived awhile CPU support isn't always great long term for early socket adopters because of changes in power requirements or other technical changes.
 
So many big releases in 2026 on brand new fab nodes:

Nova Lake - up to 52 cores on 18A
Zen 6 - up to 48 or 64 threads on N2
UDNA - merging all GPU IP into one on N2
Rubin - more performance over blackwell on N2
Someone knows when & if the stupid Lake naming will end?
 
Someone knows when & if the stupid Lake naming will end?
Who cares? Lake is just code name, like bergamo or Picasso or Rubin etc. RTX 3/4/5 is common name, just like 13/14 etc for Intel, Zen 3/5 AMD etc.
 
No need to get pedantic over exact 2:7 or 2:8 ratios, unless that data changes the underlying point, which it doesn't.

No need to become so defensive when asked for clarification on where your numbers stem from.

So for same die area they could have made 24+0+4, or gone with 16+32+4 P/E/LPE.

Emphasis mine. A third more die area is not the same.

If you made them up or did 'napkin math' that's fine, just don't present them as facts to build upon. It is consequential to have the correct information, I have a CPU Database to maintain and knowing what parts got bigger or smaller will lend to that cause, hence why I asked where you got your numbers. Pulling numbers out of our bumholes to throw around in arguments isn't what I'm here for, I'm here for facts not conjecture.

Unless that $1800 was also napkin math

Not napkin math.

Here's a refurb for even cheaper.

Unless you're implying that Intel's top consumer CPUs will go from $5-700 for decades, to significantly different pricing, which would be suicide in face of competition

Well aside from the times within the last 20 years in which Intel has, in fact, raised prices(P4 670, Q6600, i7 970, 12900KS): Intel can do whatever it wants with pricing. AMD is likely to increase prices with wafer costs going up, Intel probably will too. I remember when they used to trade places at the $1000 price point every other release. It is what it is.

You're also comparing against the base RAM M4 Max, which makes little sense to buy considering the types of task these chips excel at. Apple math means comparable RAM amounts add up to a lot more than "$1800" to get the 96/128 GB you would pair a workstation based on x86 consumer silicon with.

I am using the base configuration, 10p+4e, 36GB/512GB Studio as a comparison, correct. It's a full PC ready to use right out of the box for a similar price. You're putting up goal posts for what it "should be" while ignoring what it is. The base spec Studio is perfectly adequate mid-upper tier consumer computer for a lot of light workstation use, just like the average Ultra 9 285K PC. Your statement, I'll remind you, was that the M4 Max was a significantly more expensive platform full stop. If you want to pay more you absolutely can, but you do not have to.

I'll be unfollowing this thread now because this spiraled far out of control. I simply wanted one clarification and to point out that items A and B could be had for the same cost, but somehow that had to become a full blown argument. Why does every interaction with you have to become such a battle? So exhausting.
 
Intel has gone insane with core counts. Can't imagine what these would be used for as very few tasks scale CPUs cores up like this. I feel like 8P + 8E + 4LPE is the sweet spot for any high end PC outside datacenter stuff. Oh well, here is to hoping they are more competitive to AMD next time. They reigned in their power consumption at least.
I think we're all glad Intel, AMD and everyone else who makes a CPU, doesn't think like you suggest.
 
tbh, im only interested in this to see how AMD will react.
AMD steppig up from the 16 core part for consumer after 4 generations would not be a bad thing.

Consumer grade 32 core would be pretty sweet and imagine if having full X3D cache as well.
 
What’s the source of the rumors ?
The same source that came up with the 8p+32e Arrow Lake story. The fact is that it didn't happen and couldn't happen because of the 12-stop ring bus, and as a result, it also has a devastating latency, so what's up with the previous generation. So everyone who falls for this 16p+32e FOMO is just like that :kookoo:
 
Last edited:
tbh, im only interested in this to see how AMD will react.
AMD steppig up from the 16 core part for consumer after 4 generations would not be a bad thing.

Consumer grade 32 core would be pretty sweet and imagine if having full X3D cache as well.
The rumor is Zen 6 or Zen 7 will have a hybrid core approach, with full power cores, "C" cores and efficiency cores.
Even when sockets have lived awhile CPU support isn't always great long term for early socket adopters because of changes in power requirements or other technical changes.
That still isn't a good reason why, I'm sure Intel could design a long lasting socket if they wanted to, but of course it's more profitable for them and their board partners to change the socket with every new generation.
 
No need to become so defensive when asked for clarification on where your numbers stem from.



Emphasis mine. A third more die area is not the same.

If you made them up or did 'napkin math' that's fine, just don't present them as facts to build upon. It is consequential to have the correct information, I have a CPU Database to maintain and knowing what parts got bigger or smaller will lend to that cause, hence why I asked where you got your numbers. Pulling numbers out of our bumholes to throw around in arguments isn't what I'm here for, I'm here for facts not conjecture.



Not napkin math.

Here's a refurb for even cheaper.



Well aside from the times within the last 20 years in which Intel has, in fact, raised prices(P4 670, Q6600, i7 970, 12900KS): Intel can do whatever it wants with pricing. AMD is likely to increase prices with wafer costs going up, Intel probably will too. I remember when they used to trade places at the $1000 price point every other release. It is what it is.



I am using the base configuration, 10p+4e, 36GB/512GB Studio as a comparison, correct. It's a full PC ready to use right out of the box for a similar price. You're putting up goal posts for what it "should be" while ignoring what it is. The base spec Studio is perfectly adequate mid-upper tier consumer computer for a lot of light workstation use, just like the average Ultra 9 285K PC. Your statement, I'll remind you, was that the M4 Max was a significantly more expensive platform full stop. If you want to pay more you absolutely can, but you do not have to.

I'll be unfollowing this thread now because this spiraled far out of control. I simply wanted one clarification and to point out that items A and B could be had for the same cost, but somehow that had to become a full blown argument. Why does every interaction with you have to become such a battle? So exhausting.
I'll treat the passive aggressiveness with humor.

Reseller deals close to the end of a product life cycle, used units instead of straight from Apple. Comparing specific extreme edition limited SKUs, stretching it with edge cases, as usual, to pick fault with me, in particular. You know I mentioned Raptor Lake P/E cores as the 1:4, something even Intel did in their PRs, but you are continuing to pretend the 30% larger die area of a specific generation is the only relevant factual info when there's been several generations of heterogeneous architecture. Talking about 50+ core CPUs in the thread about them, as if buyers would pair them with 32 GB of RAM, then changing goalposts to current 285K. Whatever dude.

I won't continue as it's unprofessional for staff to seem to be in conflict, and you've apparently unfollowed the thread, not that I consider this even slightly stressful or "exhausting".

Have a great day.
 
I'm sure Intel could design a long lasting socket if they wanted to, but of course it's more profitable for them and their board partners to change the socket with every new generation.
I’ll just leave this here for you.

 
The rumor is Zen 6 or Zen 7 will have a hybrid core approach, with full power cores, "C" cores and efficiency cores.

That still isn't a good reason why, I'm sure Intel could design a long lasting socket if they wanted to, but of course it's more profitable for them and their board partners to change the socket with every new generation.
There were three generations of CPU supporting two generations of DDR, plus more recent rebrands on LGA-1700. During this time core counts jumped massively and ST perf went up around 30%.

Zen 5 already has hybrid SKUs with normal and C CCDs.
 
There were three generations of CPU supporting two generations of DDR, plus more recent rebrands on LGA-1700. During this time core counts jumped massively and ST perf went up around 30%.
All 3 were still Alder Lake based, and only E core counts increased. Single thread went up due to core clocks being pushed up which we know how that went for 13th and 14th gen, lol.
 
All 3 were still Alder Lake based, and only E core counts increased. Single thread went up due to core clocks being pushed up which we know how that went for 13th and 14th gen, lol.
What cpu are you using?
 
All 3 were still Alder Lake based, and only E core counts increased. Single thread went up due to core clocks being pushed up which we know how that went for 13th and 14th gen, lol.
Single core went up due to *Raptor Lake (not Alder Lake) having notably increased per core cache, a better memory controller, a refined process allowing better V/F curve leading to higher clocks, and entirely new dies with many hardware level tweaks being introduced. The same way Zen 3 is "based" on Zen 2 if we want to play the game of pretend.
 
The rumor is Zen 6 or Zen 7 will have a hybrid core approach, with full power cores, "C" cores and efficiency cores.

That still isn't a good reason why, I'm sure Intel could design a long lasting socket if they wanted to, but of course it's more profitable for them and their board partners to change the socket with every new generation.
I guess, depends how well old BIOS are maintained and all that. It is nice that AMD has keep sockets longer and it really helped when AMD was behind or far behind on performance.

If there is so little improvement going forward as we have seen from both companies like in the last few years there are other tech changes i will care about that will require a new MB when I get a new CPU.
 
Back
Top