9900ks 5 ghz all cores

[trents]

I know someone started a thread on this recently but can't find it not.

Found a nice article discussing the TDP issues that will come along with this chip binned at that speed.

https://wccftech.com/intel-core-i9-9900ks-5-ghz-cpu-127w-tdp-october-launch/

The article is drawing a parallel between the 5 ghz all core 9900ks and the infamous AMD FX-9xxx series. He posits that there will likely be only a handful of motherboards able to handle it and it will need higher end AIO water.

 

[bigtallanddopey]

For me as well the article shows that even if intel are competitive with the cpu price when compared to a similar performing or even lower performing Ryzen. It doesn’t really matter when you are spending $400 on a motherboard and another $200 for a decent cooler to cope with all that power and heat.

Of course people will lay out for all that so maybe I’m the idiot for thinking it’s all too much money.

Intel are desperate for those 10nm parts (if they exist somewhere)

 

[wingman99]

i7 9900k still has trubo boost. So I could run it on my low cost Gigabyte Z370 HD3. The only thing it would not run at 5.0GHz for me is prime95, gaming would run 5.0GHz fine.

 

[trents]

Yeah, and you'd probably see your house lights dim when running that baby under load.

 

[Woomack]

A lot depends on luck if we talk about max heat so I wouldn't say that in all cases AIO can handle this CPU under full load on all cores. For example, my [email protected] on all cores 1.3V was heating up so much that 360 AIO couldn't keep it below 95°C and on other high series, coolers was passing 100°C. There are users who say they have stable 5.0 or 5.1GHz and temps around ~90°C max, but most just complain at too high temps, or they can't see thermal throttling so it's fine when they can't see the problem.

9900K has 95W TDP in specs. In real it's much higher as CPU wattage is around 250W at stock settings. Coolers designed for 200W can't handle the 9900K. The main problem is that the heat is concentrated on a small spot (because of a small die). For me, design like that is a total fail. It's easier to cool down a large die like that in Skylake-X. In real CPUs with 140W TDP are as hard to keep at reasonable temps as 9900K which supposed to have 95W TDP.

TDP is a value that should suggest what cooler is required as a minimum to use for the CPU. Since 200W coolers can barely handle the 9900K then good luck with the 9900KS. When I say can barely handle then I mean without trottling.
AMD is not much better as their 65W TDP chips in real can reach 200W so it's maybe not translated directly into heat but still much above the declared value.

 

[mackerel]

It seems no matter how often it is described, people still fail to understand TDP does not equal maximum power output. Intel aren't alone, AMD are also affected but it is rarely mentioned. 65W TDP Zen 2 CPUs have a default value of 88W PPT which is easy to reach.

Based on my 8086k, with "only" 75% the cores the KS has, heat wasn't a problem with a Noctua slapped on it and overclocked to 5 GHz. More cores means more power and more heat, but I think there is sufficient headroom for it not to be a major problem.

 

[EarthDog]

The article is drawing a parallel between the 5 ghz all core 9900ks and the infamous AMD FX-9xxx series.

For some reason, I am annoyed by this. I mean, I get the resemblance, but even at the time, that AMD CPU was still a potato compared to any similar c/t count Intel around it. In this case, the raised boost clock (of 300 MHz Intel versus 700 MHz on that AMD) makes this hands down the fastest 8c/16t CPU around, period. With binning and such, it should be plenty doable with appropriate cooling..... the same as all overclockers use to reach 5 GHz all c/t without binning. The big difference here is Intel went from "95W" to "127W" (~33% increase on paper - I know, I know... see Mack's post) versus that thing went from 125W to 220W (a 76% increase) and was a hot and polished turd.

Yeah, and you'd probably see your house lights dim when running that baby under load.

LOL, love it! But being serious, if this happens due to a CPU, you have issues in your house, not with the processor.

 

[Mandrake4565]

For some reason, I am annoyed by this. I mean, I get the resemblance, but even at the time, that AMD CPU was still a potato compared to any similar c/t count Intel around it. In this case, the raised boost clock (of 300 MHz Intel versus 700 MHz on that AMD) makes this hands down the fastest 8c/16t CPU around, period. With binning and such, it should be plenty doable with appropriate cooling..... the same as all overclockers use to reach 5 GHz all c/t without binning. The big difference here is Intel went from "95W" to "127W" (~33% increase on paper - I know, I know... see Mack's post) versus that thing went from 125W to 220W (a 76% increase) and was a hot and polished turd.

I agree ED, one cannot really compare the Fx 9xxx chips to the Intel 9900. Those Fx 9xxx chips were also no where near 220w, maybe 300w and I'm being generous.

 

[Woomack]

I agree ED, one cannot really compare the Fx 9xxx chips to the Intel 9900. Those Fx 9xxx chips were also no where near 220w, maybe 300w and I'm being generous.

8120 after some OC was overloading my single stage cooler which was tuned for 300W and it wasn't even FX 9k which was released at 220W TDP ... I remember that you had even better OC results using higher thermal capacity custom water cooling than I had on ss

 

[Zerileous]

In terms of cooling these things, I think thermal conductivity from the the die into the cooling solution is a bigger issue than total capacity. The best transfer is going to happen with the lowest cold plate temp. Where as capacity would reflect the amount of thermal energy the cooling system can handle overall. With soldered IHS even being inadequate it may require some innovative ideas to improve transfer of heat our of the CPU. It would be interesting to do an evaluation of cold plate temps with a NHD15, CLC/AIO and custom loop with excess rad capacity to see if there is a significant difference.

 

[EarthDog]

Im with you. The problem is the transistor density and size of the die getting the heat out.

 

[Alaric]

The hazards of core counts. When the cooling problems begin to involve getting the heat into the cooler more than the cooler dissipating said heat, it's time to rethink the die size. It's not like there isn't enough room on a mobo for a slightly larger socket. I realize making everything smaller is supposed to be progress, but it has to be in the context of the whole.

Shrinking the heat source without lowering the output (even raising it!) will hit a wall relatively quickly. Do you want plasma cutters? Because that's how you get plasma cutters. LOL

 

[wingman99]

Im with you. The problem is the transistor density and size of the die getting the heat out.

I don't agree. It is the increase of clock speeds creating all the extra heat. Link https://www.overclockers.com/forums...s-effect-on-temperature?p=8127933#post8127933