intel core i9-14900kf idle temp 70C and 1.55V in BIOS normal?

Remove the heatsink and verify the thermal interface imprint on both CPU and heatsink looks good.

If it does, clean off and apply same way and same amount you did before. Make sure there's no plastic film (usually for shipping) still installed, and that any other protection for shipping (sometimes a light coat of oil) have also been removed.

Are these idle temps at the OS reporting idle load (typically averaging less than 1% CPU, some spikes are okay) and both CPU voltage and CPU clocks down at expected idle values? IIRC idle clocks should be between 800mhz and 1100mhz. I forget typical idle voltage range but 1.55v sounds high, do you have power saving states enabled?

IIRC on the latest UEFI you should be able to select an Intel baseline power profile or something (I forget the exact terminologies).


View: https://www.reddit.com/r/ASUSROG/comments/1cbfbbg/rog_maximus_z790_hero_how_to_set_intel_baseline/


Finally that is a very small heatsink for an i9-14900KF, I would expect idle temps to be okay but not load.

You have a 250W CPU and an 88W heatsink. That might not be the problem but it isn't helping, either.

The Asus firmware is not great and there's no telling what state the CPU is in when it enters the config screen. 1.5V is what the CPU wants for full power, 6GHz operation. If the CPU is idling at that voltage there is something wrong. Try flashing the latest BIOS, then "load optimized defaults", reboot, then enable the Intel baseline configs and reboot again.

Buildzoid looked into the crashing that recent Intel flagships are doing, and found the crashing is probably caused by voltage droop from whatever weird load-line settings the mobo bios picks, and that the ‘Intel baseline’ in the newest bioses that’s supposed to fix it just blasts the voltage way past sanity, on top of the mobos generally pushing higher voltages already.

TLDR: set a max current to something like intel’s ark’s listed VCC for your model, such as 300A-ish for 14900k. Or, change the load-line setting to turbo or extreme.


View: https://m.youtube.com/watch?v=jNwFFJyAqQU

misterlamide said:

It’s on a asus tuf gaming z790-plus motherboard

Click to expand...

Uh oh...

and I’m using the Noctua NH-D9L which has a 80mm fan

Click to expand...

The NH-D9L comes with one 92mm fan.

Before you do anything else, I recommend a contact frame to replace the integrated loading mechanism (ILM). Contact frames evenly distribute the clamping force across the entire integrated heat spreader, mitigating the bending that the ILM causes on Socket 1700 CPU IHSes.

That cooler is entirely unsuitable for the 14900KF unless you drastically limit power in the BIOS. It only has "medium" turbo headroom according to Noctua's own website.

You're on a 8P+16E 125W CPU. Intel's datasheet (page 98) lists PL1 as 125W, PL2 as 253W, and Tau as 56 seconds. I don't see a D9L tolerating 253W for 28 seconds, let alone 56 seconds.

Asus has the BIOS manual here. Disable MultiCore Enhancement in the BIOS, then drill down and set Long Duration Package Power Limit to 150, Package Power Time Window to 56, and Short Duration Package Power Limit to 150. If the CPU continues to thermally throttle, drop the Long Duration Package Power Limit to 125W. You can keep lowering the Package Power Limit settings until you get to a point that the CPU no longer thermally throttles.

As an example, I have a Thermalright Silver Soul 135 on a 12600K, in a MSI Z690I Unify motherboard. MSI's default is to set both power limits to 288W and Tau to 56 seconds. This causes my CPU to thermally throttle as the Silver Soul 135 cannot handle a ~190W load, let alone 288W. The actual spec for a 12600K is PL1 125W, PL2 150W, and Tau 56 seconds. I set PL1 and PL2 to 150W in the BIOS (Intel said starting with Alder Lake that PL1 can be set equal to PL2 if there's adequate cooling) and the cooler now handles the heat load just fine.

With a 14900KF, I wouldn't go with anything less than a large dual tower air cooler rated for 250W. Ideally, you move to a good 280mm or 360mm AIO like one from Arctic.

misterlamide said:

Is there any air cooler that supports this thing? I see Noctua has dual 120MM fan Heatsink, would that do the job?

Noctua NH-U12A​

Click to expand...

Noctua claims that their U12A can tame the beast. From my limited experience with that particular cooler I can say that it is either the best or very near the best cooling you can get from a humble little 120mm single tower.

It won't be able to achieve Noctua's traditional strength, though. I can't imagine that 250+ watt CPUs can be cooled silently by any normal means.

What specific case? The U12A isn't that small. You could potentially go with something larger that is the same height.

GN had a roundup for 2023. Might be something in there.

Worst-case, you can limit the CPU to 200W in the BIOS and that should tame it when paired with the U12A.

The other issue is you need to get the heat out of the case. If you don't have good case fans, changing the cooler will have minimal impact.

I think you may have the wrong CPU for your needs. An HTPC in a media cabinet is not the right environment (at least from my imagined version of the media cabinet) for a crazy high thermal load CPU. I would see if you can return it and step down to a 14700 or similar, maybe lower. Unless you have very important reasons for wanting that CPU it is almost certainly overkill for most media center type situations.

Alternately, you need a lot of airflow in/through the cabinet as well as the improved heatsink and fan, and the right fan design for the case to optimize airflow. That case was not designed for the kind of high thermal load you are dumping into it. The power supply and case fans will need to be setup right to improve air flow in and out of the case and the cabinet needs a good supply of cool air and a good exhaust unless it is literally just wire shelving and no enclosure.

I would have gone with a separate NAS for your drives to get them out of the media center area, if possible. Much easier to get a basic NAS with 2, 4 or 5 drive bays that are all hot swap and good system efficiency. Then make the machine a more slim/efficient design in a modern case that gets the heat out more effectively without a ton of fan noise. That big case is basically a big hot box unless it has effective fans and that can mean more noise.

Yeah, as @Paladin is saying, you've chosen a very bad chip for your purpose. The 14900 burns an assload of heat, it's crazy hot compared to prior generations, and you're trying to use it as an HTPC, which typically doesn't need much horsepower. Further, you bought an F chip, which doesn't have onboard video. That's often exactly wrong for an HTPC, because much of the software in that space can take advantage of the onboard video acceleration functions. That can also use the acceleration on a discrete card, but you can often get away without a separate GPU these days, saving electricity and heat. Not having a GPU, however, does put additional heat load on the CPU, which could be problematic in your specific case.

You'll probably want to turn the CPU heat limits way, way down. You'll gimp the chip in terms of throughput, but you usually don't need much CPU throughput in an HTPC. Set PL1 and PL2 to like 75 watts, and it should stop overheating and run okay.

A much smaller and cheaper chip, with onboard video, would have been a better match for your intended use. You're kinda-sorta trying to use a V8 to drive a go-kart.

edit: also, as @continuum and @IceStorm are suggesting, checking to be sure you removed all film, have good heatsink compound contact, and have a contact frame will help. Even a chip that monstrous should idle really cool. Idling at 70C indicates a cooling problem.

Where you should be seeing things melt down is when you put a load on it. Idle should be good no matter what chip you're using.

If it is getting 1.55V the cooler interface is not the problem. Running at 70º is in fact quite respectable for that voltage state.

jwbaker said:

If it is getting 1.55V the cooler interface is not the problem. Running at 70º is in fact quite respectable for that voltage state.

Click to expand...

Even at idle? I haven't used an Intel chip in quite awhile, but I thought one of their big advantages was idling at like 5W?

The reverse has been a major criticism of Zen 3, that the chips idle hot.

malor said:

Even at idle? I haven't used an Intel chip in quite awhile, but I thought one of their big advantages was idling at like 5W?

The reverse has been a major criticism of Zen 3, that the chips idle hot.

Click to expand...

"At idle" would be a lower voltage. If it was idle it would have the VRM programmed for .780V. That's how P-states, i.e. Turbo, work. 1.55V is P0, the highest possible power state.