[SOLVED] 9900k temps getting too hot (+90C) while XMP is activated (under Prime95 small ffts Non AVX)

Sep 2, 2020
10
0
10
0
Hey there, I have got a 9900k on stock settings with 2x8 RAM 3000mhz overclocked with XMP, every time i try prime95 small ffts non avx temps tend to go between 90 to 95 degrees (tried with and without XMP), everytime I try with AVX it shoots up to 100 degrees; while gaming temps tend to go maximum maximum between 60s - 70s, discover some peaks to 76 (maximum) with HWMonitor.
After that gaming session I tried again small ffts with and without XMP, with it it gets to 98 degress (probably would go to 100 as I stopped it 1 minute later) without XMP it stayed at 95 maximum.
Just being in windows temps stay between 30-40s.
While using davinci resolve I rendered a 4k 60 fps video and it got to maximum 93 degrees.
Everytime I remove the glass side panel temps tend to get some degrees lower.

Besides XMP everything else on BIOS is on auto or so (i have not changed anything else).

NOTE: this is the third 9900k that has been on my hands and has shown the same problem (first 2 of em went through an z390 e gaming from asus, this last one only on a "just testing" gigabyte gaming X (it was lend to me while i wait my meg z390 ace to come (as tech service gave me warranty for this stuff).
NOTE 2: Lots of BSOD while I had my first 9900k and the z390e, since the 2nd and 3rd 9900k not even one BSOD.
NOTE 3: While being on asus strix one i discovered that when XMP gets activated, my cpu voltages went up, but while being on the gaming x cpu voltages just... stay the same.
Build:
9900k NON OC (i dont really want to do it yet (And i dont really need to...)
z390 gaming X (had a rog strix z390e, waiting a meg z390 ace)
the most sad thing_ coolermaster ma610p (only 1 heatsink) which i dont have enought money to swap to like a noctua... but have also heard noctua temp issues; AIO gets me scared... and on tech service they tried a 240 mm one and temps went higger too..
rtx 2080 super.
and so..

What is concerning me is that this thing is not even overclocked!, and you may say "CPU has HW errors.." but this is the third one! jhaha... dont know if prime 95 just gives unrealistic stress levels...
 

vov4ik_il

Respectable
Undervolted -0,05, BSOD, now my m.2 ssd is gone, is not being detected
That's' pity. For SSD, check your UEFI settings, it might have something to do with it being reset after unsuccessful undervolt.
What utility did you use? I used Intel Extreme Tuning Utility. I have another system with 8700K and also have it undervolted by -0.1v. Undervolting brings temps down significantly, you will literally see watts dropping in the utility itself once you hit apply...
Here is what I get with 9900 (non-k):


To elaborate on the role of voltage in this deal, check the Ohm's Law wheel below. The resistance (R) of the chip at full load is nearly constant and it goes up when it is idling. Voltage is a pre-programmed curve (and depends on frequency) if it is dynamic or a set value if it is static.
Lowering the value brings power consumption down exponentially. All power draw of your chip turns into heat.

 
Last edited:
What is concerning me is that this thing is not even overclocked!
Doesn't mean that it's running intel stock settings,mobo makers often overvolt the CPUs both on intel and amd platforms.
Since 10th gen has a huge range that it can work in it can be pushed really high by the mobo.
dont know if prime 95 just gives unrealistic stress levels...
Well unrealistic for normal mainstream users it's very specialized, most software will have much lower numbers.


Also glass side makes me think you got a case that is more style than function, open case should normally give you higher temps since the air flow depends on a semi closed system.

Also also, have you tried the bios fan settings? If it's set on quiet it will provide the least amount of cooling to prevent throttling which means the highest temps.
 
Looking at HWMonitor, what clock speeds are sustained on all cores during yourPrime95/small FFTS/no AVX workload? (That will help determine if you are truly running 'stock', where 4.7 GHz should be the max all-core turbo speeds encountered...; if you see 5.0 GHz, then MCE is enabled, and should be disabled, given your current temps)
 
Sep 2, 2020
10
0
10
0
Doesn't mean that it's running intel stock settings,mobo makers often overvolt the CPUs both on intel and amd platforms.
Since 10th gen has a huge range that it can work in it can be pushed really high by the mobo.

Well unrealistic for normal mainstream users it's very specialized, most software will have much lower numbers.


Also glass side makes me think you got a case that is more style than function, open case should normally give you higher temps since the air flow depends on a semi closed system.

Also also, have you tried the bios fan settings? If it's set on quiet it will provide the least amount of cooling to prevent throttling which means the highest temps.
How do I know if im on really stock settings?

I got a... agh is a really unknown one but it is a Gamemax Abyss TR case, it has an infinity display on front and a glass pannel on one of its sides.

Bios fans settings are:
CPU FAN SPEED CONTROL : normal
Fan control use temperature input: CPU
Temperature interval: 1
CPU fan control mode: Auto
CPU FAN STOP: disabled
 
Sep 2, 2020
10
0
10
0
Looking at HWMonitor, what clock speeds are sustained on all cores during yourPrime95/small FFTS/no AVX workload? (That will help determine if you are truly running 'stock', where 4.7 GHz should be the max all-core turbo speeds encountered...; if you see 5.0 GHz, then MCE is enabled, and should be disabled, given your current temps)
Well. it is weird as it maintains on 4.7, but! I just ran another small ffts and in one moment it showed me a max peak of 5.0, I restarted HWMonitor and now the maximum is 4703, and in another moment it climbed to 4.8
In another moment one of the cores climbed to 4.9. I simply dont know
EDIT: Yeah definetely, he is sometimes climbing to 5.0

Now im running another test and temps aer on the 98 maximum and stable at 95
 
Last edited:
Sep 2, 2020
10
0
10
0
Also, i dont know what is he doing but it is like staying at 4.7 but out of nowhere he is getting doing to 800MHz for 1 second then climbing up again
 

Cheese Dripper

Prominent
Jan 5, 2020
37
3
545
5
2 things, 1: make sure airflow is good in your case. And 2: Make sure the cooler has been attached properly with enough thermal compound. Also make sure you use an AIO as the 9900k can get quite hot and air cooling may not be sufficient.
 
Sep 2, 2020
10
0
10
0
2 things, 1: make sure airflow is good in your case. And 2: Make sure the cooler has been attached properly with enough thermal compound. Also make sure you use an AIO as the 9900k can get quite hot and air cooling may not be sufficient.
I opened the glasspannel and even put a normal fan outside and... yeah... nearly 10 degrees lower... would a noctua nh d15s work?
 

vov4ik_il

Respectable
Undervolted -0,05, BSOD, now my m.2 ssd is gone, is not being detected
That's' pity. For SSD, check your UEFI settings, it might have something to do with it being reset after unsuccessful undervolt.
What utility did you use? I used Intel Extreme Tuning Utility. I have another system with 8700K and also have it undervolted by -0.1v. Undervolting brings temps down significantly, you will literally see watts dropping in the utility itself once you hit apply...
Here is what I get with 9900 (non-k):


To elaborate on the role of voltage in this deal, check the Ohm's Law wheel below. The resistance (R) of the chip at full load is nearly constant and it goes up when it is idling. Voltage is a pre-programmed curve (and depends on frequency) if it is dynamic or a set value if it is static.
Lowering the value brings power consumption down exponentially. All power draw of your chip turns into heat.

 
Last edited:
Aug 31, 2020
19
0
10
0
That's' pity. For SSD, check your UEFI settings, it might have something to do with it being reset after unsuccessful undervolt.
What utility did you use? I used Intel Extreme Tuning Utility. I have another system with 8700K and also have it undervolted by -0.1v. Undervolting brings temps down significantly, you will literally see watts dropping in the utility itself once you hit apply...
Here is what I get with 9900 (non-k):


To elaborate on the role of voltage in this deal, check the Ohm's Law wheel below. The resistance (R) of the chip at full load is nearly constant and it goes up when it is idling. Voltage is a pre-programmed curve (and depends on frequency) if it is dynamic or a set value if it is static.
Lowering the value brings power consumption down exponentially. All power draw of your chip turns into heat.

I used BIOS, but I undervolted directly Vcore from 1200 to 1190. whats the difference between that and vcore offset?

That BSOD i got from undervolting didnt damage anything... did it?
 

vov4ik_il

Respectable
I used BIOS, but I undervolted directly Vcore from 1200 to 1190. whats the difference between that and vcore offset?
If you set a fixed voltage, you are missing a lot of power-saving that is given to you by the dynamic curve. In other words, the power draw does not drop as low on low usage and frequency as it would with a curve. I would set the voltage to Dynamic or Auto (depends on your BIOS) and only use voltage offset from the curve.

That BSOD i got from undervolting didnt damage anything... did it?
Might cause data corruption, just like it would in case of a power interruption or abnormal shutdown, but unlikely. Hardware damage would only (unlikely) be possible if there is a higher voltage applied (overvoltage), not your case. Unless you adjusted PCIe bus voltage it has nothing to do with your SSD.
 
Last edited:
Sep 2, 2020
10
0
10
0
If you set a fixed voltage, you are missing a lot of power-saving that is given to you by the dynamic curve. In other words, the power draw does not drop on low usage and frequency as it would with a curve. I would set the voltage to Dynamic or Auto (depends on your BIOS) and only use voltage offset from the curve.


Might cause data corruption, just like it would in case of a power interruption or abnormal shutdown, but unlikely. Hardware damage would only (unlikely) be possible if there is a higher voltage applied (overvoltage), not your case. Unless you adjusted PCIe bus voltage it has nothing to do with your SSD.
Okay thank you!! I will try later... but at the end I just ordered a Noctua nhd15, Just got tired of not getting the performance I paid for just by having that obnoxious coolermaster cooler. I know it is cooler related as this is the Third CPU on a different MB, this should do the trick. And maybe I will even be able to Overclock it a little
 

vov4ik_il

Respectable
Okay thank you!! I will try later... but at the end I just ordered a Noctua nhd15, Just got tired of not getting the performance I paid for just by having that obnoxious coolermaster cooler. I know it is cooler related as this is the Third CPU on a different MB, this should do the trick. And maybe I will even be able to Overclock it a little
For my 8700k I used 280mm AIO and de-lidded it to get temps under control. I am skeptical about the cooler you ordered, but it is just me. It does look better than the one you currently use.
Keep us posted :)

P.S. Found my old post (prior to delid) where I attach the undervolting and prime95 with AVX load screenshot with temps.
 
Last edited:
Sep 2, 2020
10
0
10
0
For my 8700k I used 280mm AIO and de-lidded it to get temps under control. I am skeptical about the cooler you ordered, but it is just me. It does look better than the one you currently use.
Keep us posted :)

P.S. Found my old post (prior to delid) where I attach the undervolting and prime95 with AVX load screenshot with temps.
Just got my noctua, had to remove side glass panel as the ram would not allow me to have both fans without elevating one of em, So, the situation is this, without AVX (and with or without XMP, temps are the same) and without glass panel and both fans i get roughly 77-81 temps, (The fans are not even at 100% speed as PWM reduces it at that temperature) Now, if i activate AVX i get 90-93 degrees, nothing more than that...
Without that cooler i usually got with avx instantly 100 degrees, without avx I instantly got 94-95, so we are talking about a 15 degrees difference.
Now, without not doing anything, just being afk, temps are 35-37 with fans not spinning at full speed, if i put full speed temps lower to 29-30 (which i have seen is a nice temperature, taking into account that where i live is a tropical country so... well temperatures tend to change a lot but my days are always 18 degrees or higher).

When I try removing the lateral fan that has conflicts with my ram temps go 3-5 degrees higher, with glass panel temps get up to... the same as before.. so this is a chassis problem... and also, i have a better Cooling system now so it wont be a big deal. Im just waiting to change my chassis, but rn, i can handle without having a glass panel, i can always just cover it with something (not clothes) I have prepared for it when i dont use it.

Then I just gotta acept that 9900k is a hot cpu, and that i would nearly ever see temps like that (while normal using temps stay in 60s, and nearly never get to 70s, I think that while playing (lot of stuff to do so havent been able to try) temps will probably go to 70-75 maximum, and if i use a headset and put those fans to full power i think ill be good
 

ASK THE COMMUNITY

TRENDING THREADS