Question My cpu runs at 90 C under load please help

Aug 30, 2020
8
1
15
0
CPU: i9-9900k
GPU: 2080ti gaming x trio
CPU cooler: NZXT x63
MB: MSI mpg z390 gaming edge ac
RAM: 32gbTeam T-group Delta RGB 3200mhz
PSU: Thermaltake 850w smart pro RGB
temps reach 90-95 C under max load and sit around 45-55 C idle
please help I just got the computer and I'm very scared
 

mdd1963

Champion
Is that under an MCE-enabled load (all-cores at 5 GHz), or, under Intel-intended load of all cores at 4.7 GHz? IF at 5 GHz all core turbo, likely quite typical....

Check your temps with Prime95, small FFTs (not smallest), and AVX/AVX2 disabled..(even those temps will typically be 8-10C above those ever seen while gaming...)

As for the various 240-280mm AIOs, not will handle the sustained high clock speed/MCE-enabled heat produced by a heavy all-core at 5 GHz rendering load...
 
Although the 9900k can be overclocked you need a top 360mm AIO rad if your target frequency is 5GHz.
Are you Overclocked?. If so then bring back frequency to default.

The 9900k has a tjmax of 100C so you are safe at 90C for short periods. It is sustained high temps that can harm the CPU.

Voltage on the core is what determines temperature so what is your Vcore in Bios?
You may have more voltage than you need for stability so try decreasing Vcore till the system is no longer stable. Drop voltage in .01V steps and stress test each step till the system no longer boots. Then raise Core voltage in .01 till stable.
This will be the best system settings for safe operation under load conditions.
 

Desch_

Reputable
Apr 29, 2016
104
13
4,545
1
if you just got the computer maybe it wasnt stored in ideal conditions and the thermal paste is totally dried out...idk just a thought, thats where id start.

also, what software ru using to determine the temps?
 
Aug 30, 2020
8
1
15
0
Is that under an MCE-enabled load (all-cores at 5 GHz), or, under Intel-intended load of all cores at 4.7 GHz? IF at 5 GHz all core turbo, likely quite typical....

Check your temps with Prime95, small FFTs (not smallest), and AVX/AVX2 disabled..(even those temps will typically be 8-10C above those ever seen while gaming...)

As for the various 240-280mm AIOs, not will handle the sustained high clock speed/MCE-enabled heat produced by a heavy all-core at 5 GHz rendering load...
I was running all cores at 5ghz at first, but then I reset the bios back to default.
Where can I change the prime95, small ffts, and the avx/avx2?
 
Aug 30, 2020
8
1
15
0
Although the 9900k can be overclocked you need a top 360mm AIO rad if your target frequency is 5GHz.
Are you Overclocked?. If so then bring back frequency to default.

The 9900k has a tjmax of 100C so you are safe at 90C for short periods. It is sustained high temps that can harm the CPU.

Voltage on the core is what determines temperature so what is your Vcore in Bios?
You may have more voltage than you need for stability so try decreasing Vcore till the system is no longer stable. Drop voltage in .01V steps and stress test each step till the system no longer boots. Then raise Core voltage in .01 till stable.
This will be the best system settings for safe operation under load conditions.
Yea that's what I did, and I did that for a while but the problem didn't go away.
My bios settings are currently at default, but I'll try and see if that works.
BTW, what do you mean by "stable?"
 
Aug 30, 2020
8
1
15
0
if you just got the computer maybe it wasnt stored in ideal conditions and the thermal paste is totally dried out...idk just a thought, thats where id start.

also, what software ru using to determine the temps?
I built the computer 3 weeks ago so I'm not sure what you mean by not ideal conditions
I use NZXT CAM to monitor the temps.
 
"Stable" means the system runs successfully under load.

Quote/
Where can I change the prime95, small ffts, and the avx/avx2? End quote/

Don't use Prime95 avx for stress testing. If a worker stops it will not tell you why. A better product is AIDA64. It's free and you can stress test the CPU, FPU and cache. It will also test your GPU and RAM.
 

Desch_

Reputable
Apr 29, 2016
104
13
4,545
1
I built the computer 3 weeks ago so I'm not sure what you mean by not ideal conditions
I use NZXT CAM to monitor the temps.

ahh your original post read "i just got the computer..."

so i assumed you bought a prebuilt.

meaning it could have sat in a warehouse for months or longers before you got it. If their climate control went out for a day or two it could cause problems for anything with thermal paste in there. or if it was on a no refrigerated truck for a couple days dirving thru the desert in 110F heat, that could have compromised the thermal paste as well. Sorry, misunderstood!

Did you use the factory paste that comes on the CPU cooler or did you wipe that clean and use fresh out the tube, high quality, afermarket paste? because the issues i mentioned above could have easily happened to the thermal paste that comes standard on your heatsink.

I always wipe that grey garbage off with alcohol wipes and add nice new silver based thermal paste. New thermal paste can easily bring temps down by 10c, and 80c under load is a lot more comfortable of a ride than 90c....
 
Aug 30, 2020
8
1
15
0
also a quick reddit search says nzxt cam can give inaccurate readings for a lot of people.

i would install cpuid and speccy and check temps on those applications as well. maybe youre getting a misreading.

https://www.guru99.com/best-cpu-temperature-monitor.html

they are number 1 and number 3 on this site, and just the ones im most familiar with, so thats why im reccomending. they are freeware.
interesting! I used the CUPID HWMonitor and the temperatures shows to be at 83 peak.
Tho my idle temps are still pretty high.
I will try using a better thermal paste and see if that makes a difference
 
Reactions: Desch_
Aug 30, 2020
8
1
15
0
"Stable" means the system runs successfully under load.

Quote/
Where can I change the prime95, small ffts, and the avx/avx2? End quote/

Don't use Prime95 avx for stress testing. If a worker stops it will not tell you why. A better product is AIDA64. It's free and you can stress test the CPU, FPU and cache. It will also test your GPU and RAM.
Which cpu test should I run? there are a lot of tests and options but I really don't understand them.
 
Go to the Tools section and select the System Stability test. Tick the check boxes for CPU. FPU and Cache and run the stress test for 10Mins Initially. As the test progresses keep an eye on your temps in the graphs. I also have HWinfo64 along side Aida64 to also monitor rail voltages and as a double check of temperatures at the die.
The other tests can be run separately.
If you decide to replace your TIM then use either AS5. MX4 or Thermal grizzly which are good. Use it sparingly as the paste is there to fill just the tiniest of voids.
 
Last edited:
Aug 30, 2020
8
1
15
0
Go to the Tools section and select the System Stability test. Tick the check boxes for CPU. FPU and Cache and run the stress test for 10Mins Initially. As the test progresses keep an eye on your temps in the graphs. I also have HWinfo64 along side Aida64 to also monitor rail voltages and as a double check of temperatures at the die.
The other tests can be run separately.
If you decide to replace your TIM then use either AS5. MX4 or Thermal grizzly which are good. Use it sparingly as the paste is there to fill just the tiniest of voids.
And when running the tests, what temps should my CPU be at? I ran it for about 10 min and using the HWMonitor, I observed pretty high temperatures. (https://ibb.co/rbhj4SJ)
Also, my idle temps are at 50 C and I really don't know why
 
Last edited:
Yes those temps are a little high under load however you would rarely see that under normal conditions.
High 90C temps if sustained will degrade the CPU over time.
For longevity I like to keep below 80C under load and 10-15C above ambient room temp at idle.
 
Aug 30, 2020
8
1
15
0
Yes those temps are a little high under load however you would rarely see that under normal conditions.
High 90C temps if sustained will degrade the CPU over time.
For longevity I like to keep below 80C under load and 10-15C above ambient room temp at idle.
So are there any good ways to lower my temps?
 

mdd1963

Champion
"Stable" means the system runs successfully under load.

Quote/
Where can I change the prime95, small ffts, and the avx/avx2? End quote/

Don't use Prime95 avx for stress testing. If a worker stops it will not tell you why. A better product is AIDA64. It's free and you can stress test the CPU, FPU and cache. It will also test your GPU and RAM.
If a worker crashes, one can start with the conclusion that said rig is unstable with Prime95... :)

Small FFTs w/AVX enabled is an overload, however....the defacto standard of '100% TDP equivalent load'with the newest P95 version is to disable AVX and AVX2, small FFTs.... (To poster who asked where to change options, the Prime 95 options (and small/smallest FFTs, AVX /AVX2 enabled) come up whenever you launch the Prime95 application. The default load is 'Blended', which is not as strenuous a load as small FFTs, typically with sustained temps in Blended mode about 7-9C lower than the temps induced by small FFTs.
 
Thanks for the quote mdd1963 and you are correct however they are opinions as are mine.

In saying that, it is my opinion and that of many others that P95 is a torture test alright. It is a harsh test that should be left to chip testers and just one of many algorithms for stress testing.
It's like putting your foot to the floor of your car and leaving it there. Some torture their CPU overnight only to find their system crashes after 5mins of game play. More than likely as the result of a diminished CPU.
Whether you use P95 or any stress test imo you should NEVER leave the system unattended.

In the end you do what you feel is the best way and the argument is OLD.
I use AIDA64 with good results for initial testing however the best test is a Real World tester and that is "Realbench" and If you can pass that test then you know your stable.
 
Last edited:
Reactions: Afro_ninja199

ASK THE COMMUNITY

TRENDING THREADS