News Core i9 14900KF Breaks World Record, Almost Achieves 9.1GHz

[Admin]

With today's arrival of Intel's new Raptor Lake Refresh chips, a team of overclockers has already broken a world record with Intel's new i9-14900KF flagship, achieving a 9043.92 MHz CPU frequency world record run.

Core i9 14900KF Breaks World Record, Almost Achieves 9.1GHz : Read more

 

[Alvar "Miles" Udell]

CPU Overclocking is considered a sport now?

 

[ezst036]

CPU Overclocking is considered a sport now?

I could see it. If you want to see bizarre, check out hobby horse competition.

 

[cyrusfox]

CPU Overclocking is considered a sport now?

I mean slap boxing is a thing, why not?

 

[derekullo]

How are any minute gains huge?
Wouldn't that be ... huge gains?
Huge gains for whom? the electricity/helium producing companies?
If this article wasn't AI created I think AI could have helped lol!

 

[spongiemaster]

How are any minute gains huge?

It's the difference between absolute and relative.

.11 seconds is a pretty minute amount of time, however that was how much Usain broke his previous world record in the 100M which was a huge margin. The previous 5 records combined dropped the 100M record by .10 seconds.

 

[bit_user]

The team was able to achieve a very impressive 9043.92GHz on a single P-core with liquid helium, breaking the previous world record by 35.1MHz.

Heh, that was obviously meant to say "9043.92 MHz".

It's also funny how the article points out the significance of a mere 35.1 MHz, but then takes the liberty of rounding up their achieved speed by a whopping 56.08 Mhz!
🤣

WCCFTech did the same thing. I'm not sure who copied off of whom, but I think they both rip stuff from each other, these days.

 

[bit_user]

CPU Overclocking is considered a sport now?

For many years, actually. They've been holding organized competitions for over a decade.

Toms once had a champion overclocker writing some articles for the site. I'll see if I can find them, but the site's search feature is so bad I'll probably have to resort to Google.

 

[spongiemaster]

Heh, that was obviously meant to say "9043.92 MHz".

It's also funny how the article points out the significance of a mere 35.1 MHz, but then takes the liberty of rounding up their achieved speed by a whopping 56.08 Mhz!
🤣

WCCFTech did the same thing. I'm not sure who copied off of whom, but I think they both rip stuff from each other, these days.

If you read the article, you would see that the overclockers were able to boot the system at 9.1Ghz but it wasn't stable enough to get a verification of the clock speed for an official record.

 

[bit_user]

If you read the article, you would see that the overclockers were able to boot the system at 9.1Ghz but it wasn't stable enough to get a verification of the clock speed for an official record.

How often is that the case, though? If validation didn't matter, it wouldn't be required.

 

[Order 66]

10GHz is going to happen sooner rather than later. It's too bad Intel's prediction didn't come true.

 

[bit_user]

10GHz is going to happen sooner rather than later. It's too bad Intel's prediction didn't come true.

You mean this?

"Realistically speaking, we should be able to see NetBurst based processors reach somewhere between 8 – 10GHz in the next five years before the architecture is replaced yet again."

The future of Intel's manufacturing processes

www.anandtech.com

He goes on to say:

"what would you possibly want to do with a 10GHz processor?"

--Anand Lal Shimpi (Dec, 2000)
​

We also have to put that in some perspective: it's not a Golden Cove @ 10 GHz, but a Netburst core. Golden Cove is easily faster at like 5 GHz.

 

[bit_user]

Toms once had a champion overclocker writing some articles for the site. I'll see if I can find them, but the site's search feature is so bad I'll probably have to resort to Google.

Ah, I remembered the name: Splave

Here's one of his more recent articles:

Core i9-12900KS World Record: How I Hit 7.8 GHz on 8 Cores

70 liters of liquid nitrogen works wonders.

www.tomshardware.com

Here's a brief bio and his entire list of articles:

Articles by: Allen 'Splave' Golibersuch | Tom's Hardware

Discover the latest content written by Allen 'Splave' Golibersuch, author at Tom's Hardware.

www.tomshardware.com

 

[deesider]

How often is that the case, though? If validation didn't matter, it wouldn't be required.

Speed records must be run in both directions - they need to zip a file, then unzip it

 

[HideOut]

it did NOT almost break 9.1ghz. Your math fails. It broke 9ghz, but in order to almost break 9.1 it'd need to be near 9.1. So over 9.05 you could make that argument. But let me guess, somewhere in that story there is a link that the Future plc gets a cut of a sale from so it sounds like a good eye catching story to publish.

 

[Alvar "Miles" Udell]

So 23 years after this infamous statement

Realistically speaking, we should be able to see NetBurst based processors reach somewhere between 8 – 10GHz in the next five years before the architecture is replaced yet again.

Intel has sort of delivered on that.

 

[Geef]

But... Can it run Crysis?

 

[Sluggotg]

It was fun back when we could overclock our Celerons, (the 300a), from 300mhz to 450mhz with a simple jumper setting and it ran flawlessly. Times have changed a bit but we all benefit.

 

[bit_user]

So 23 years after this infamous statement

Intel has sort of delivered on that.

Well, the statement was talking about stock speeds, so I guess that's what you mean by "sort of". I actually wonder whether CPUs will ever reach 10 GHz, stock.

I also wonder what would happen if they ported the Netburst microarchitecture to the Intel 7 node. What would it naturally clock at? And you could probably fit like 1000 Northwood cores on a die the size of Sapphire Rapids'.

 

[Order 66]

Well, the statement was talking about stock speeds, so I guess that's what you mean by "sort of". I actually wonder whether CPUs will ever reach 10 GHz, stock.

I also wonder what would happen if they ported the Netburst microarchitecture to the Intel 7 node. What would it naturally clock at? And you could probably fit like 1000 Northwood cores on a die the size of Sapphire Rapids'.

I would love it if intel would do that, or at least run some kind of simulation (I know it’s ridiculous, but bear with me) to see how many cores and at what clocks they could get with it.

 

[Alvar "Miles" Udell]

I actually wonder whether CPUs will ever reach 10 GHz, stock.

It depends. If CPU speed ratings were based on processing capabilities in relation to a fixed standard, then they'd be well over 10ghz by now compared to then (since IPC improvements have increased drastically since then), but since that's highly improbable, I don't see 10ghz stock CPUs until silicon-on-insulator is replaced with new materials, since we haven't seen a substantial increase in clock speed in the last 20 years, since the Pentium 4 landed with 3ghz base clock speeds we haven't hit 5ghz yet (think the highest was 4.7ghz with the FX-9590), with 5ghz+ being "turbo" or overclocking. The increased power draw and heat production just increases instability too much.

 

[bit_user]

It depends. If CPU speed ratings were based on processing capabilities in relation to a fixed standard

No, I'm strictly talking about clock speed.

we haven't seen a substantial increase in clock speed in the last 20 years, since the Pentium 4 landed with 3ghz base clock speeds we haven't hit 5ghz yet (think the highest was 4.7ghz with the FX-9590), with 5ghz+ being "turbo" or overclocking.

If we're talking about trends, then there's been an undeniable upward trend, ever since the "reset" Intel did with Core 2. In that case, it's more informative to look at median, single-core clock speeds than either base (which is counteracted too much by increasing core counts) or the high watermarks.

As for where "turbo" fits into this... I guess I'd say we should eliminate time-limited boosting from the data, and just look at sustained, single-core clock speeds.

 

[Order 66]

No, I'm strictly talking about clock speed.


If we're talking about trends, then there's been an undeniable upward trend, ever since the "reset" Intel did with Core 2. In that case, it's more informative to look at median, single-core clock speeds than either base (which is counteracted too much by increasing core counts) or the high watermarks.

As for where "turbo" fits into this... I guess I'd say we should eliminate time-limited boosting from the data, and just look at sustained, single-core clock speeds.

7600x is also 4.7ghz. If AMD continues in the direction of high base clocks with zen 5 we may get 5 GHz base clocks.

 

[bit_user]

I would love it if intel would do that, or at least run some kind of simulation (I know it’s ridiculous, but bear with me) to see how many cores and at what clocks they could get with it.

Not only that, but I have long wondered such things about even simpler cores.

The closest Intel got to this was the first Xeon Phi, which was allegedly made by taking Pentium "54C" cores and updating them to have both 512-bit vector extensions (these were slightly different than AVX-512) and 4-way hyperthreading. They were able to fit up to 61 active cores per die, using a 22 nm process node, back in 2012-2013.

Xeon Phi - Wikipedia

en.wikipedia.org

At the time, the largest conventional Xeon Intel made on 22 nm had 15 active cores.

Intel Ivy Bridge–based Xeon microprocessors - Wikipedia

en.wikipedia.org

 

[drtweak]

And yet only recently has the AMD FX record been broken. I miss my 8320. Was rock solid for a long time. Mobo is what gave out.