Intel 14th Gen Core K-series Specs Leaked

[T0@st]

Benchlife claims to have obtained full specifications of Intel's upcoming 14th Gen Core series—the site kicked things off by releasing details of a trio of Raptor Lake Refresh K-series SKUs earlier today. Insiders have seemingly divulged fairly comprehensive specs for i9-14900K, i7-14700K, and i5-14600K desktop CPUs. The expected lineup-wide implementation of greater clock speeds (+200 MHz) is present on these examples according to the leaked info—i9-14900K is reportedly capable of boosting up to 6.0 GHz (via Thermal Velocity tech), while its Core i7 and Core i5 siblings are said to be hitting 5.6 GHz and 5.3 GHz (respectively).

The Core i7-14700K seems to be the only rumored model to receive a core count increase—the listed 8P+12E configuration is decked out with more Gracemont efficiency cores when compared to the 13th Gen equivalent's makeup (i7-13700K, 8P+8E). This grants a slightly increased pool of Intel's "Smart Cache"—33 MB instead of the previous gen model's 30 MB. These 125 W TDP "K" SKUs are expected to arrive mid-October alongside "KF" models (lacking iGPUs). The 65-W non-K lineup could be presented at the next CES, and launched in January 2024.




VideoCardz has collated the leaked information, and has updated its comparison charts (referencing 12 & 13th Generation Core SKUs):



View at TechPowerUp Main Site | Source

 

[lemonadesoda]

Can P and E cores run with asymmetric clocks? Because if they are designed for multitasking, and I steer non-time-essential background apps/threads to E cores, I don't want them speeding up and making more heat and consumption just because I have an active fast-to-complete thread running on my P cores. What my P-cores superboosting, and my E cores running cool with high efficiency.

===========

Should they have delayed the announcement of LGA-1851? Did they basically just say, use Core iX-14 as an upgrade path for socket 1700, but if you are hanging out for a new platform, socket 1700 is end of the road, so wait for 1851? Oh wait. It's Intel. They expect you to upgrade your platform for every tick-tock iteration of x86!

 

[dgianstefani]

Can P and E cores run with asymmetric clocks? Because if they are designed for multitasking, and I steer non-time-essential background apps/threads to E cores, I don't want them speeding up and making more heat and consumption just because I have an active fast-to-complete thread running on my P cores. What my P-cores superboosting, and my E cores running cool with high efficiency.

===========

Should they have delayed the announcement of LGA-1851? Did they basically just say, use Core iX-14 as an upgrade path for socket 1700, but if you are hanging out for a new platform, socket 1700 is end of the road, so wait for 1851? Oh wait. It's Intel. They expect you to upgrade your platform for every tick-tock iteration of x86!

This is the 3rd CPU family to release on LGA 1700, so I'm not sure what you're complaining about.

 

[Nostras]

I'm especially interested in the 14100. If the price remains the same it's quite the budget offering.
But knowing Intel it's going to be an Alder Lake chip with kneecapped clocks.

14600K looks really weak compared to the 14400 and 14500. Wonder if it's once again Raptor lake only 14600K and Alder lake 14400/14500.
Else there's not much reason to go 14600K...

 

[bonehead123]

I don't understand why they don't just go back to the +, ++, +++, +++++ etc naming scheme, cause this "lake.refresh.refresh "crap is annoying AF, hehehehe

 

[docnorth]

At least if they can further improve efficiency, like from Alder Lake to Raptor Lake (efficiency, not consumption), it will be interesting. I’m not very optimistic though, my 13700 non-K reached almost 300w without limits.

 

[dgianstefani]

At least if they can further improve efficiency, like from Alder Lake to Raptor Lake (efficiency, not consumption), it will be interesting. I’m not very optimistic though, my 13700 non-K reached almost 300w without limits.

It's no different to any of the main CPU manufacturers.

For example, X3D chips aren't inherently more efficient than the normal X chips, they're just voltage/power capped to a sane number because of the cache intolerance to the voltages the chips normally run at.

Nothing stops you from setting a 150 W or a static voltage limit and seeing a whole 5% lower performance if the power draw bothers you.

There should be some efficiency improvements as the process is further matured and all of the SKUs should be based on the Raptor lake core not the Alder lake core, which has a larger cache and other improvements.

 

[lemonadesoda]

This is the 3rd CPU family to release on LGA 1700, so I'm not sure what you're complaining about.

Oh I'm not complaining.

It's just a demonstration of uncoordinated PR. Day n they release last Core i socket 1700. This has taken months of work and planning. But on day n-1 other PR team announce socket 1851. If you can't see the stupidity in that order of PR then you are overlooking the obvious. Big company stovepipes.

 

[Denver]

It should be forbidden to release old products under a new name, and worse make it look like a "new generation".

 

[Deleted member 232534]

 

[docnorth]

It's no different to any of the main CPU manufacturers.

For example, X3D chips aren't inherently more efficient than the normal X chips, they're just voltage/power capped to a sane number because of the cache intolerance to the voltages the chips normally run at.

Nothing stops you from setting a 150 W or a static voltage limit and seeing a whole 5% lower performance if the power draw bothers you.

There should be some efficiency improvements as the process is further matured and all of the SKUs should be based on the Raptor lake core not the Alder lake core, which has a larger cache and other improvements.

I agree and planned anyway to set a power limit at circa 180, although the Noctua nh-u14s can handle the temporary 200w limit. But first I’ll try to undervolt and see if a lower limit is needed. I’m not complaining at all and feel ‘rewarded’ because idle, low load and gaming consumption are very low.

 

[ZoneDymo]

It should be forbidden to release old products under a new name, and worse make it look like a "new generation".

yeah, its just a words war but It would be nice if this was Alderlake V(ersion) or Rev(ision) 3.

 

[john_]

Intel can't/wouldn't add more P cores. Not enough room with a 10nm++++ and at the same time the marketing advantages of advertising a big number of cores.

The question is, what about this? Users of Hybrid CPUs can verify or not what is shown in the video

This is the 3rd CPU family to release on LGA 1700, so I'm not sure what you're complaining about.

Is it? I mean, if AMD was offering only quad core Zen1 CPUs for the Ryzen 1000 series, then was adding six cores models to call it Ryzen 2000 and then eight cores models and call it Ryzen 3000 and finally was making the Zen1+ refresh and calling it Ryzen 4000, would that be 4 series or one, maybe two? Taking for example AM4, we had in fact 3 series, based on Zen 1(1000 and 2000), Zen 2(3000 and 4000) and Zen 3(5000 and X3D). Zen 1+, the Ryzen 2000 series was just a refresh it's questionable if we should count it.

On LGA 1700 we have 12th series and 13th series where the main difference is with P cores, with E cores being the same architecture and with 14th gen we have a refresh where we get in fact different models of the same architectures, not new architectures, if I am not mistaken. So this, in best case scenario, is two CPU families, not 3. The 3rd is just the 2nd getting more family members and a new name. And considering half the 13th gen is still Alder Lake models, maybe it's more like one and a half series, not two.

 

[Denver]

yeah, its just a words war but It would be nice if this was Alderlake V(ersion) or Rev(ision) 3.

Well, It would be less absurd if intel just released a KS version of the i9 and i7 models or something like that.

 

[THU31]

The question is, what about this? Users of Hybrid CPUs can verify or not what is shown in the video

I saw this video yesterday and it's the first time I'm hearing of this.

I haven't experienced any of these issues in the two months since I upgraded. But I have E-cores and all power saving features disabled.

It doesn't make sense that it would only affect some users. And it's been almost two years since the release of Alder Lake.


On topic, disappointed if the 14600K turns out to just have extra 200 MHz with 6 P-cores. That's basically an irrelevant change.

 

[trsttte]

Good grief, 16 e-cores is just excessive. How many background tasks are there supposed to use all that!?

They're going all in after the marketing points of achieving a better cinebench score even if users and workloads hate the configuration

 

[lemonadesoda]

Good grief, 16 e-cores is just excessive. How many background tasks are there supposed to use all that!?

…extra e-cores so all the new telemetry doesn’t lag the PC so much.

 

[john_]

Good grief, 16 e-cores is just excessive. How many background tasks are there supposed to use all that!?

They're going all in after the marketing points of achieving a better cinebench score even if users and workloads hate the configuration

It's not about background tasks, it's not about Cinebench. It's about number of cores advertised. And 24 are more than the competition's 16.

 

[Minus Infinity]

So Videocardz leaked specs from last week are already changed for the 14600. It was 8P+8E last week, now it's back to 6P+8E. I guess that would have been too good to be true.

Hopefully all the lower tier models get upgraded to Raptor cove and Gracemont+ cores at least.

Finally is DLVR inlcuded or not. Some rumours claiming it will use less power at same clocks as 13th gen.

 

[Od1sseas]

Good grief, 16 e-cores is just excessive. How many background tasks are there supposed to use all that!?

They're going all in after the marketing points of achieving a better cinebench score even if users and workloads hate the configuration

When will people realize that E-Cores aren't just for background tasks. They significantly boost performance in EVERYTHING except gaming like much faster rendering.
If E-Cores where only for background tasks and Cinebench, the i9 13900K will have got TRASHED by the 7950X in productivity because it would have been 8 cores vs 16. Jesus, people those days.

 

[Wirko]

Finally is DLVR inlcuded or not. Some rumours claiming it will use less power at same clocks as 13th gen.

It may never be relevant for desktop CPUs. As I understand it (and I could be wrong), it is effective at idle and at low loads.

 

[Denver]

When will people realize that E-Cores aren't just for background tasks. They significantly boost performance in EVERYTHING except gaming like much faster rendering.
If E-Cores where only for background tasks and Cinebench, the i9 13900K will have got TRASHED by the 7950X in productivity because it would have been 8 cores vs 16. Jesus, people those days.

Yeah, because "EVERYTHING" is perfectly optimized to work on MT. That must be why the best selling CPUs are the 5800x3D and 7800x3D, and even the 5600.

 

[Od1sseas]

Yeah, because "EVERYTHING" is perfectly optimized to work on MT. That must be why the best selling CPUs are the 5800x3D and 7800x3D, and even the 5600.

The majority of apps are perfectly optimized for MT, yes. That's why you get better performance the more cores you add (For example going from 8P+8E to 8P+16E). Watch a benchmark. There are multiple on the internet.

 

[chrcoluk]

I think the bigger news would be if cache has increased rather than clock speeds, it did from 12th to 13th so might do again.

Can P and E cores run with asymmetric clocks? Because if they are designed for multitasking, and I steer non-time-essential background apps/threads to E cores, I don't want them speeding up and making more heat and consumption just because I have an active fast-to-complete thread running on my P cores. What my P-cores superboosting, and my E cores running cool with high efficiency.

===========

Should they have delayed the announcement of LGA-1851? Did they basically just say, use Core iX-14 as an upgrade path for socket 1700, but if you are hanging out for a new platform, socket 1700 is end of the road, so wait for 1851? Oh wait. It's Intel. They expect you to upgrade your platform for every tick-tock iteration of x86!

Yes, the core clock is based on individual core load. So they dont all rise up and down in tandem unless its full on load on all cores. In the situation you described, I would also expect at least a few p-cores to be parked (not represented on the clocks reported by hwinfo, so in the screenshots below some may well be parked).

Second shot will have some parked p-cores as preferred core only gets 5.4ghz when is enough parked.

 

[Wirko]

I think the bigger news would be if cache has increased rather than clock speeds, it did from 12th to 13th so might do again.

Nah. Intel wouldn't forget to give these chips a new name if they were actually new chips.