[SOLVED] i9-10850k reaching 94C ?

ganiy.imran

Reputable
Jan 2, 2018
22
0
4,510
So I have this setup:

Lian Li Cool II Case (2 fans)

i9 10850k
Dark Rock Pro 4 CPU Cooler
Rog-Strix Z590-E Motherboard


It's been about a year that I've left it this way, but I'm wondering if I should fix it:
During a stress tests, and some games, the CPU with get pretty hot, like 80-88, and occasionally jump to 90-94

Is there something I should to improve it? Is the Dark Rock not good enough or should I try to redo the thermal paste?

If It's fine this way I'd rather just leave it then mess with it, but if this is clear indication the Dark Rock is malfunctioning I'd like to do something.
I heard this CPU runs hot and has a TJ Max of 105, so maybe it's fine?

Thanks for any input
 
Solution
Thanks for any input
Spikes aren't a problem. It's when you have it literally sitting up there for several hours straight on a regular basis that it becomes a problem.
TJ Max for the cpu is 100C, not 105C.
Dark Rock Pro 4 should be enough for that cpu, but there's probably a few other things happening all at the same time, so it's not solely the cooler at fault.
1)Airflow.
Lian Li Cool II Case (2 fans)
Not enough air entering and exiting when the gaming gets going. Power consumption goes up, air movement also needs to go up, assuming everything else is stable or operating normally.

2)Auto voltage.
Motherboards automatically apply more Vcore than needed by default, but it was worse than usual on Z590 boards. Some...
So I have this setup:

Lian Li Cool II Case (2 fans)

i9 10850k
Dark Rock Pro 4 CPU Cooler
Rog-Strix Z590-E Motherboard


It's been about a year that I've left it this way, but I'm wondering if I should fix it:
During a stress tests, and some games, the CPU with get pretty hot, like 80-88, and occasionally jump to 90-94

Is there something I should to improve it? Is the Dark Rock not good enough or should I try to redo the thermal paste?

If It's fine this way I'd rather just leave it then mess with it, but if this is clear indication the Dark Rock is malfunctioning I'd like to do something.
I heard this CPU runs hot and has a TJ Max of 105, so maybe it's fine?

Thanks for any input
I have a 10850K as well on a 240mm Asus LC240 Strix AIO, they are generally hot CPU's. You do seem a little toasty, are you running at 4.8Ghz all core or do you have something like multi core enhancement enabled and running at 5Ghz all core?
Also what's your CPU vcore, sometimes motherboards can give them a little too much.

I actually run mine at 4.5Ghz instead of 4.8Ghz as that chops a significant chunk of the power consumption off so I get only 65C full load. However I know that's considered devil worship around here.
 

Phaaze88

Titan
Ambassador
Thanks for any input
Spikes aren't a problem. It's when you have it literally sitting up there for several hours straight on a regular basis that it becomes a problem.
TJ Max for the cpu is 100C, not 105C.
Dark Rock Pro 4 should be enough for that cpu, but there's probably a few other things happening all at the same time, so it's not solely the cooler at fault.
1)Airflow.
Lian Li Cool II Case (2 fans)
Not enough air entering and exiting when the gaming gets going. Power consumption goes up, air movement also needs to go up, assuming everything else is stable or operating normally.

2)Auto voltage.
Motherboards automatically apply more Vcore than needed by default, but it was worse than usual on Z590 boards. Some options:
-Checking for bios updates that 'improve stability or performance'. They may contain improved voltage curves. If you're not comfortable with updating bios, you don't have to.
-Undervolting the cpu, which can be done either in bios or with Throttlestop. Ex: Finding Cpu Vcore setting in bios, changing it to Offset mode, and applying an offset of -0.050v. Might be able to apply -0.080, -0.100, or even higher offsets. The last 3 will require some time on your end to be sure they're ok.

3)Overclocking.
MCE, or Multi Core Enhancement. Disable it in bios, if it isn't already. It's pretty much an E-Z OC, but that excessive voltage from #2 is still there.
 
Solution

sonofjesse

Distinguished
Turn the AC down in the room, or if its winter open a window.

I would say don't worry about the stress test, monitor tips on your "heavy cpu" tasks and I bet you won't see 90C that is my guess. People will stress test things for 24 hours and say I hit 95C. Well what do you do day to day, I check email and Facebook, well you don't' have to worry about ever hitting that 95C doing them tasks lol.

That is like saying you was in vegas desert driving 140 mph for 8 hours, and your car over heated.........where do you live, KY, do you drive over 100 for more than 8 hours in 125 degree heat ? Nope, then I doubt your car will over heat lol.
 

ganiy.imran

Reputable
Jan 2, 2018
22
0
4,510
Thanks for all the replies,
My OS Is Windows 10 Home 10.0.19042, latest bios software
Low ambient room temperature; CPU Idles around 35

what I'm getting by reading what everyone wrote is:

  1. watch ambient temperature (low)
  2. under volt the CPU (currently runs anywhere
  3. update motherboard bios
  4. disable overclocking
  5. increase number of fans in the case

I also have a RTX 3080 GPU installed close the the Dark Rock cooler, maybe I should move it down a slot?

Does it seem like I installed the Dark Rock cooler correctly with the thermal paste, or should I take a look at it and do it again?
 

Phaaze88

Titan
Ambassador
I also have a RTX 3080 GPU installed close the the Dark Rock cooler, maybe I should move it down a slot?
No. Doing so doesn't really change anything.

Does it seem like I installed the Dark Rock cooler correctly with the thermal paste, or should I take a look at it and do it again?
1)Yes.
2)You're more than welcome to do it if you want. I think it's the lack of air movement, specifically when the gpu is active, and the aggressive auto Vcore.
 
A DarkRock air cooler is insufficient for a 10850K....

If it is all you have, you can try undervolting a little. and, cranking down multiplier selection to a couple hundred MHz over what would normally be utilized. (Inte's XTU will assist with this easily)
 

ganiy.imran

Reputable
Jan 2, 2018
22
0
4,510
So I installed a new set of 5 chassis fans, 1800rpm
I tested the most intense app I have, Cyberpunk 2077 with max settings, ray tracing, 1440p
It played around 75-85C with occasional brief spikes to 89.
I guess the next step is messing with the voltage, or maybe it's fine the way it is?
It seems improved with the new chassis fans.
 

ganiy.imran

Reputable
Jan 2, 2018
22
0
4,510
89C is still a little on the warm side, do you know what the all core frequency is under load and what your load vcore is.

So I'm using Ai Suite 3, the Dual Intelligent Processor section has tools for altering the CPU voltage, ratio, and CPU cache ratio.
The vcore is set to "Adaptive", and while running prime 95 small FFT's the voltage seems to go up to 1.2, but i've also seen it jump up to 1.35 under less load.

I tried setting it off adaptive and sticking it at 1.215, but then then the core speed never went above 4.0mhz, and it is normally at 4.8, the temps weren't much better in prime 95.

What can I do?

I also noticed my 1800rpm fans never reach their full speed, is this because a slower fan is connected to the same hub and it's throttling all of them?

Thanks for your assistance
 
I have a 10850K as well, mine at stock with all cores at 4.8Ghz was just under 1.35v, so that sounds roughly correct though perhaps slightly higher. Not a huge fan of Prime95, I use Aida64 and the Intel Diagnostic Tool for stress testing, however I find Chinebench the most representative for the load temps I get during normal use. To be clear I found mine to be very hot at 4.8Ghz too so I run at 4.5Ghz. Cinebench load temp is 65C, Aida64 is 71C. Though I think your cooler might be better than mine.

I'm on an AIO, but I need to run my fans hard to keep the temps down. Are you saying that your CPU cooler fans are plugged into a fan hub? If so then I wouldn't do that because the CPU fan control won't work properly, your Asus software might let you control the fans based on the CPU temp like that but I would really recommend plugging them directly into the CPU header and setting a fan curve in the bios. If you can't plug both fan's in then you'll need a splitter but I would have thought the cooler would have come with one.
 

Phaaze88

Titan
Ambassador
@Nighthawk117
What's your motherboard? The auto voltage curves of Z590 were found to be more aggressive than 400 series.
You and the OP would both benefit from undervolts.

The AIO will cool better than the DRP4 under heavy all core loads, but one needs to be able to tolerate the greater fan noise. If not, the gap between the 2 coolers shrinks unless one over-provisions on size.

There's other factors at play too, but motherboard might be one of the biggest difference here.
 
I have an MSI Z490 Mag Tomahawk. I only have a 240mm AIO not a 360mm, that's why I thought a DRP4 might be better. I actually disabled turbo boost on mine, left voltages on auto and set multiplier to 4.5Ghz, I know that's like devil worship but I prefer the cooler computer than the 6.6% performance loss. My max vcore is listed as 1.214 running Aida64 stress test.
 

Phaaze88

Titan
Ambassador
Yeah, you have a 400 series board after all.
The DRP4 likely gets better idle thermals - not that anyone should get hung up on that - but both coolers under a Cinebench/Aida/Occt load, the AIO should pull ahead - bar the louder fans.

Hey, if it gets to hot for your liking, and you downclock it as a solution - I wouldn't go so far as to call it devil worship... it's not that serious.
Auto Vcore is just too much for stress tests. I'd even advocate never leaving Vcore on auto if you're going to run those.
Consider going into bios and trying a negative Vcore offset of -0.050v and see what that does with Turbo Boost enabled. OP likely can take theirs even further, up to -0.100v.
 
Oh ok I didn't realise the 500 series board were more aggressive on the volts. To some enthusiasts it's super weird to get a K chip and run it in anyway below its full potential, my two cents would be the performance loss is too small to be perceivable but the dent it makes in heat and power consumption is large. If you go from 4.5 to 5.1 for example, it increases the performance by 13% but roughly doubles the power consumption.

Problem is if you take vcore off auto, then you lose the C state stuff. The negative vcore offset I much prefer, I've thought about doing it but haven't tried yet. I did try this once before on my old CPU but the PC cut out, I think the idle voltage got too low. Is there anything I need to be aware of with that, would I need to fiddle with LLC at all?
 

Phaaze88

Titan
Ambassador
Problem is if you take vcore off auto, then you lose the C state stuff.
If the cpu is run on either offset or adaptive Vcore mode, they should still honor C-states. There must be another setting or program turning it off.
I could see manual or override Vcore mode turning it off, as well as the mobo's Enhanced Turbo doing it. Enhanced Turbo is a general term for a board's 1 click auto OC, it is not the same as Turbo Boost. I don't know what Msi calls theirs.
Oh, and if you're running Dragon Center, it probably does it too.

I'm running my cpu on Adaptive Vcore mode, and C-states are working as they should; the cores not doing anything are running at 1.2ghz, with the active ones jumping to 4.6ghz(7820X).

LLC: I imagine most don't need to touch it.
 

ganiy.imran

Reputable
Jan 2, 2018
22
0
4,510
So I underclocked the CPU from 4.8 to 4.5, which caused auto voltage to decrease significantly
Replaced the chassis fans with better ones and added more.
Average temps on difficult apps are usually no more than 75, so I think it's good.
Thanks for everyone's input. It really helped
 

TRENDING THREADS