News Core i9-13900K Soars To 8.2 GHz, Leaving Ryzen 7000 In The Dust

[Admin]

Allen 'Splave' Golibersuch overclocks Intel's Core i9-13900K Raptor Lake processor to 8.2 GHz under liquid nitrogen.

Core i9-13900K Soars To 8.2 GHz, Leaving Ryzen 7000 In The Dust : Read more

 

[emike09]

I'll finally be able to run Flight Sim maxed out without a CPU bottleneck!!! Mom! More LN2!!!

 

[Colif]

Get back to me when any of this is useful or possible on air. Its rather meaningless otherwise.

 

[spentshells]

I overclocked my lapped q6600 on air to a stable 3.8Ghz. Thats 60% wake me up when the percentage gets that far.

Also though... gg im super jealous

 

[Vanderlindemedia]

Is it a suicide run? Then it's useless to even compare what the max oc'ing potential would be on air.

I overclocked my lapped q6600 on air to a stable 3.8Ghz. Thats 60% wake me up when the percentage gets that far.

Also though... gg im super jealous

Its never going to happen with this type of silicon. We are at the max of what is possible for quite some time. Things are slowing down significant.

 

[-Fran-]

That was single core and it looks like they tried 6.3Ghz "all-core" (P-cores only), but didn't record it or pass. Still cool, I guess.

I think Sandy Bridge was a better OC'er.

Regards.

 

[ocer9999]

I wonder if under the hood was a OC-Formula there...

 

[gdmaclew]

My Dad is bigger than your Dad.
Oh Please.

 

[DavidLejdar]

Yeah, it is nice to hear about overclocking, but I am certainly one of those who are interested in the out-of-box performance only. And I don't plan to do any overclocking, with the reasons against it:

• extra power consumption,
• possibly voiding warranty and/or reducing life-span (if not done properly), and
• if I would need that extra performance, I should go for a higher tier, or if I would get the highest tier, then not necessarily making full use of it. Specifically, if someone has e.g. the Odyssey Neo G8 screen, or a 360hz screen or recently even a 500hz screen, then they sure may be interested to push the FPS as far as they can get, for which some overclocking may be needed. But for me, some stable 120 frames per second (in most games) will be plenty good, and that should be achievable with out-of-box hardware.

 

[Unolocogringo]

Some of us have an overclocking addiction !!!!!
Been doing it since late/ fall of 96. when I was introduced to The original Thomas Pabst"s (hope I spelled it correct) website.
P75@90
P200@252 25%
P3 450@600 33%
P3 550e @869 63%
p3 750@ 997 33%
P31.26@1589. 12.5%
Then the P4 fiasco (to keep it family friendly). And I went to the dark side.
Now days without dry ice/ phase change/ LN2/ LHe2 those numbers are just memories.

 

[SkyBill40]

Still can't beat an AMD processor from nearly a decade ago. Interesting.

 

[ezst036]

Isn't this just below the record Vishera?

 

[Co BIY]

Why is there a MAPP gas torch next to the OC rig ?

 

[TerryLaze]

Why is there a MAPP gas torch next to the OC rig ?

Since they use LN2 the CPU goes to waaaay sub zero temps and at some point the CPU stops working, if you need to quickly get a CPU back to operating temps instead of waiting for it to thaw naturally you need a torch.

 

[spentshells]

Is it a suicide run? Then it's useless to even compare what the max oc'ing potential would be on air.




Its never going to happen with this type of silicon. We are at the max of what is possible for quite some time. Things are slowing down significant.

What interests me is seeing just the e cores overclocked.
I'd like to see the others disabled and these e cores brought up as far as they can go.

When the ecores were first released I doubted thier capability and i'm clearly wrong.

 

[spentshells]

Still can't beat an AMD processor from nearly a decade ago. Interesting.

worse yet is without an overclock they all pretty much crush it in performance.

 

[samopa]

CMIIW, isn't that RaptorLake still using 10nm and Ryzen 7 using 5nm ? How can processor that so much behind in production process can overtake the much more advanced production process ?
Is the 5nm not live up to its name or is the Intel Fabs so much superior than TSMC's ?

Somebody, enlighten me please

 

[Specter0420]

CMIIW, isn't that RaptorLake still using 10nm and Ryzen 7 using 5nm ? How can processor that so much behind in production process can overtake the much more advanced production process ?
Is the 5nm not live up to its name or is the Intel Fabs so much superior than TSMC's ?

Somebody, enlighten me please

Maybe a little of both. Intel can afford, and has, the best engineers on the planet and in great numbers.
But the "nanometer measurement" hasn't been an accurate way to measure density, they abominated it into some kind of marketing BS. So Intel 10 is closer to Ryzen's 5nm than in sounds too.

 

[-Fran-]

CMIIW, isn't that RaptorLake still using 10nm and Ryzen 7 using 5nm ? How can processor that so much behind in production process can overtake the much more advanced production process ?
Is the 5nm not live up to its name or is the Intel Fabs so much superior than TSMC's ?

Somebody, enlighten me please

To make a very light paralel, it's the same reason why a V6 can make as much power and torque as a V8. You have different ways to build the block, pistons and mixture so the engine can perform differently according to whatever metric and goals you want.

The way Intel optimises their processes (and their goals) are different to TSMCs and other foundries. Materials they use, tools (asterisk here) and tunning; all that turns into hard measures of distance of different elements/parts that, back in the day (probably above 45nm?), were not as important as they are now and actually do affect things which did not before. To name one that's very common to find information about: electro-migration or "tunnel effect" (quantum tunneling) on super low voltages.

I've always had a personal amazement to Intel's fabbing capability and I still do. I mean, we all meme about 14nm, but even with that anchor, the process actually held them for quite a while until they sorted what they called "true 10nm" (caveats, yes). Now they're trying to tweak 10nm and transition to higher density keeping the same overall metrics and goals as always, which is good.

Regards.

 

[TheOtherOne]

Get back to me when any of this is useful or possible on air. Its rather meaningless otherwise.

You know the sad truth about this is, Intel can and will advertise to mass consumers with the usual "Upto XX numbers" saying their product is that much more powerful than the competition acting like it's a reality for average users. They might have to add a tiny * at the end of that line and then a footnote in very small text that says, it "may" be possible to achieve "upto" that number using extremely high cooling methods.

They will technically be within the boundaries of consumers law when it comes to false advertisement.

 

[NatalieEGH]

Sounds really great. A10 second run I am guessing just enough to get the processor speed verified.

Of course one has the money for a nitrogen collection and compression and heat release system to keep the system running 24x7 under a steady stream. It sounds like a really cool (pun intended) exercise. I am guessing the equipment will be in the 100's of thousands of dollars if not millions. Let's face it, if Google can skip a few generations of processors and motherboards by using liquid nitrogen cooling and it was economically viable, I am sure they would.

I am not convinced running under those conditions are not harmful to the long term operation of the CPU.


I am personally hoping for the cooling in the chip processors such as those being tested.

https://blog.rittal.us/what-is-direct-to-chip-liquid-cooling

TSMC Mulls On-Chip Water-Cooling for Future High-Performance Silicon

TSMC takes its turn at the chip-cooling water fountain.

www.extremetech.com

Researchers demonstrate in-chip water cooling

Chip is designed so that water channels are next to its hottest components.

arstechnica.com

Asetek Direct to Chip (D2C) liquid cooling technology - Asetek

Asetek Direct to Chip (D2C) liquid cooling technology increases data center density and enables maximum sustained CPU throughput.

www.asetek.com

This Microchip Has Its Own Built-In Cooling System

The key to the new approach is that the cooling system has been designed into the chip right from the start to maximize heat transfer.

singularityhub.com

View: https://www.youtube.com/watch?v=Csoahf6SkuM

View: https://www.youtube.com/watch?v=YdUgHxxVZcU

https://www.facebook.com/watch/?v=377417603364095

 

[TerryLaze]

You know the sad truth about this is, Intel can and will advertise to mass consumers with the usual "Upto XX numbers" saying their product is that much more powerful than the competition acting like it's a reality for average users. They might have to add a tiny * at the end of that line and then a footnote in very small text that says, it "may" be possible to achieve "upto" that number using extremely high cooling methods.

They will technically be within the boundaries of consumers law when it comes to false advertisement.

Intel already released the info for the 13900k and all they advertise is 5.8 on a single core and 100% of CPUs will be able to hit that even with simple cooling.
Single core power draw is at like 50-60W for the 12900k,
https://www.intel.com/content/www/u...-36m-cache-up-to-5-80-ghz/specifications.html

You are confusing intel with AMD, it was them that advertised boost clocks that only 50% of the CPUs would hit, and that 50% was the best case scenario, for some skus it was only 5-10% of CPUs that managed to hit the advertised boost clocks.

View: https://youtu.be/DgSoZAdk_E8?t=576

 

[Xenx]

I overclocked my lapped q6600 on air to a stable 3.8Ghz. Thats 60% wake me up when the percentage gets that far.

Also though... gg im super jealous

Gotta remember that the way CPUs are marketed is different now. They bake OC into the CPU now. Technically, the base clock on the 13900k is 3GHZ. The 13900k all-core factory OCs to 5.5ghz. That's 183% over base clock. Beat that.. ;-)

 

[SkyBill40]

worse yet is without an overclock they all pretty much crush it in performance.

Well, yeah. One would expect that given the advances in tech and how the Bulldozer architecture wasn't the greatest. I wouldn't call that even remotely close to a "worse yet." It's a nearly decade old chip, dude. Take your L and go home.

 

[TerryLaze]

Gotta remember that the way CPUs are marketed is different now. They bake OC into the CPU now. Technically, the base clock on the 13900k is 3GHZ. The 13900k all-core factory OCs to 5.5ghz. That's 183% over base clock. Beat that.. ;-)

That's not just technically, if you can only provide 125W of power or cooling then it will only run at 3Ghz under heavy load.
anything above that is auto OC if you can provide what the CPU needs to do it.

Also that is 183% of the performance but "only" 83% over base clock.