Wednesday, October 20th 2021

Retail Version of Intel Alder Lake Core i9-12900K Overclocked to 5.2 GHz on All Cores

Although we can't confirm it, the screenshot below is said to be of a retail version of the Intel Alder Lake Core i9-12900K that has been overclocked to 5.2 GHz on all P-cores, with the E-cores at stock clocks. It is said to be drawing a massive 330 Watts at these clock speeds, which is rather a lot for a consumer level CPU.

Sadly details such as the motherboard used and RAM clocks are absent. The E-cores are said to be locked at a maximum clock of 3.7 GHz, so there appears to be no overclocking potential in them. Yes, Intel does manage to edge out AMD's Ryzen 9 5950X at these clock speeds in the multithreaded test, which is no mean feat considering we're looking at eight threads less here, but Intel does so at over twice the power draw.

Update: Updated due to a slight misunderstanding, the E-cores were apparently enabled, but running at stock clock.
Sources: bilibili, via @9550pro
Add your own comment

67 Comments on Retail Version of Intel Alder Lake Core i9-12900K Overclocked to 5.2 GHz on All Cores

#51
dicktracy
Very impressive. 8 big cores + 8 Atom cores that don't even have HT utterly destroying 16 big ones. Like Intel said "it's over."
Posted on Reply
#52
InVasMani
I wonder if the E cores can be clocked higher with BCLK and just how much leeway it has. If it's like Haswell at 102MHz BCLK that's no fun at all relative to Skylake.
Posted on Reply
#53
TheoneandonlyMrK
dicktracyVery impressive. 8 big cores + 8 Atom cores that don't even have HT utterly destroying 16 big ones. Like Intel said "it's over."
It'll be over when AMD finishes glueing on some extra Memory to those cores , and increases the tdp to half what's used here:D:)
Posted on Reply
#54
Initialised
Yawn, 2600K did 5GHz ten years ago... Which brick of the 5GHz wall would you like to headbutt next?
Posted on Reply
#55
Crackong
It would be so fun watching launch day reviews

My guesses are:

1. With Win11 = barely usable, with Win10 = Disaster
2. With DDR5 = Wins in productivity but lose in gaming due to high latency
3. With DDR4 = Ties with Zen3/11th in gaming, lose in productivity.
4. Power consumption = fallout fusion cells
Posted on Reply
#56
Metroid
my ryzen 5900x at 4.5ghz is 120w real and my aio 2 fans 240mm cooler is for 150 watts, this intel cpu would need a quad aio 480mm to function at 330watts, insane, too much power. I wonder how it would do at 120w, same power as my ryzen 5900x.
Posted on Reply
#57
fevgatos
ZoneDymoThe thing is with that argument is that that Ryzen cpu, in this case, would be the 5950x which would absolutely beat the crap out of the 12900k if it was clocked to 5.2 ghz.

I mean you are correct but also wrong, you are correct that in normal usage etc this cpu will probably be fine, im looking at getting a 12600k myself and maybe a bigger raptorlake one later, but this article is not about a normal usage scenario and we are judging it in that scenario.

Like my original post, 5.2ghz all core does not impress, having to turn off the E-cores makes it even worse and having it consume threehundredandthrity watts in the process is just the cherry on the dissappointment cake.
I'm sorry but that doesn't make ANY sense. Let me put it this way. This CPU will be absolutely incredibly fine and cool running blender at, let's say, 4.8ghz. I think you agree with that. Ok, great. The fact that you have the OPTION (that's what it is) to make it run at 5.2ghz with an increased power consumption is a PRO, cause it's your CHOICE. I wish we had the same CHOICE with the ryzen cpu's, being able to overvolt them and run at 5+ ghz. We can't.

For example, i've seen some cinebench runs on a 5950x with LN2 running at 6+ ghz. It probably consumes over 500watt at that frequency. Does that mean it's a bad CPU that consumes a lot of power? Nope. So it's the exactm same with the 12900k, only difference is you don't need exotic cooling, you can do it in your house.
TheoneandonlyMrKAny CPU sat idling, is cool.

Some of us don't Do idle though. .

So you being fine or for that matter me being fine is not everyone is fine with 330W.
Nobody said anything about idle. Running blender a 10900k at stock beats the 5800x at stock while being 30-35C degrees COOLER. You can see it in the review from this very site ;)
Posted on Reply
#58
Metroid
CrackongIt would be so fun watching launch day reviews

My guesses are:

1. With Win11 = barely usable, with Win10 = Disaster
2. With DDR5 = Wins in productivity but lose in gaming due to high latency
3. With DDR4 = Ties with Zen3/11th in gaming, lose in productivity.
4. Power consumption = fallout fusion cells
Plus, ddr5 will be insanely expensive and we know those ddr5 are the bad ones, the good ones always come after a while, remember when ddr4 came, speed was at 2133mhz at very high latencies 18cl, at moment we have 3600mhz at 18cl. This ddr5 will be no different.
Posted on Reply
#59
watzupken
dicktracyVery impressive. 8 big cores + 8 Atom cores that don't even have HT utterly destroying 16 big ones. Like Intel said "it's over."
It is highly likely that Intel will win back the single core performance consistently, but multithreaded performance may not outperform a 16 core 5950X. In addition with this kind of power draw required, chances of Intel trying to claw back sales to enterprise/ data center is not likely to happen with ARM and AMD completely dominating when it comes to performance vs power efficiency. While it is likely that Intel can lower clockspeed to drop power draw, it also means they lose single core performance.
Posted on Reply
#60
Crackong
OMG Corsair had the solution for this


Posted on Reply
#61
KarymidoN
MakaveliI'm going to assume you are referring to something like this?

However if I'm not mistaken that says Total System Power.

Only the 12900k OC = 330W
Total System Power 5990x OC = 332W

intel Fanboys = BUT AMD DRAWS A LOT OF POWER TOO YOU KNOW
Posted on Reply
#62
TheLostSwede
News Editor
sillyconjunkieOne review. One set of hardware. One set of software. One set of measurement variables. Numerous other variables. I'm not trying to pick a fight but we all tend to think our way of doing it is best. Regardless of test variables, the performance per watt can't just be ignored.

Other examples of all core overclocks show a range of 272 watts to 360 watts depending on test type and lots of other stuff. Call their data disingenuous and I'll stand by my statement that the performance per watt for 8 vs 16 cores is impressive. Banned? Probably imminent..


skatterbencher.com/2021/08/30/skatterbencher-27-amd-ryzen-9-5950x-overclocked-to-5117-mhz/

hothardware.com/reviews/amd-ryzen-5000-zen-3-processor-review?page=4

www.kitguru.net/components/cpu/luke-hill/amd-ryzen-9-5950x-zen-3-cpu-review/17/
Posted on Reply
#65
ThrashZone
Hi,
I doubt I'll bother with this release
A) boards will be stupid priced just like z590 boards are
B) 12900k will also be way too much at launch
But I contacted optimus and asked if they had a mounting bracket for socket 1700
They said prior foundation and sigV2 water blocks are compatible, which I find hard to believe but okay :cool:
Posted on Reply
#66
Makaveli
ThrashZoneHi,
I doubt I'll bother with this release
A) boards will be stupid priced just like z590 boards are
B) 12900k will also be way too much at launch
But I contacted optimus and asked if they had a mounting bracket for socket 1700
They said prior foundation and sigV2 water blocks are compatible, which I find hard to believe but okay :cool:
You forgot DDR5 will be overpriced because its new.
Posted on Reply
#67
elgato
So what is the opinion about air cooling at stock clock?
I was considering a Noctua nh-d15 in a high airflow case.

Do you think it will be significantly different than a 5900X?
Posted on Reply
Add your own comment
Jun 1st, 2024 15:06 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts