Question Another Ryzen 5800x3d Temps topic

biL4

Reputable
Feb 27, 2019
19
1
4,515
Hi!

Just managed to put my build together. After a bad experience with a faulty MSI PSU got my hands on a new one and started my new pc.

ASUS ROG STRIX B550-A GAMING - Latest Bios Version 2803 2022/04/29
Ryzen 5800x3d
Arctic Freezer 420 ARGB
Mushkin Redline Lumina RGB DDR4 32GB 4000MHz CL18 Dual Kit
Western Digital SN850X 1TB
Sharkoon SilentStorm Cool Zero 850W Gold
Lian Li Lancool III
Crap GPU for video signal :)

Default bios setting excepting docp/xmp or whatever is called


Idle temps at 33-35C
Cinebench R21 ~75C cpu boost under 4.2 GHz
Cinebench R21 ~88 cpu boost 4225-4250
Prime95 goes directly at 90C

Max cpu fan speed

I saw that for this CPU 90C is something normal in Prime95 but I was expecting lower temps on cinebench with this cooler. Tried normal mount and offset and also reapplied thermal paste for each time. Also connected the cooler to the motherboard as it comes out the box (single 4pin in cpu header or pump and currently is mounted with fans connected to cpu header and pump to pump header)
Each time the results were the same.

Anyone with a similar cooler or maybe a less capable one that is having better temps or lower?

Thnx
 
Let us know how that goes.

Yes, that will do it for the paste. The idea as you know, is to spread it evenly and thinly across the entire IHS.

Doe the CPU throttle when it's hits the higher temps?

My 5600x hits about 70c on Cinebench R23, with a 4.65 allcore. So, your temps aren't much higher than that.

have you tried a voltage offset on the CPU?
 

biL4

Reputable
Feb 27, 2019
19
1
4,515
Let us know how that goes.

Yes, that will do it for the paste. The idea as you know, is to spread it evenly and thinly across the entire IHS.

Doe the CPU throttle when it's hits the higher temps?

My 5600x hits about 70c on Cinebench R23, with a 4.65 allcore. So, your temps aren't much higher than that.

have you tried a voltage offset on the CPU?
Its kind of weird because when running Cinebench R21, at first the clock does not go over 4.2GHz (~75C) but after some runs its boosts to 4225 and the temps goes up to ~88C

Regarding voltage offset on the CPU, I have to look in to that....

edit
Went down a bios version and still hit 88C but higher clock
 
Last edited:
Its kind of weird because when running Cinebench R21, at first the clock does not go over 4.2GHz (~75C) but after some runs its boosts to 4225 and the temps goes up to ~88C

Regarding voltage offset on the CPU, I have to look in to that....

edit
Went down a bios version and still hit 88C but higher clock

But you're clearing CMOS after each bios update? Yes?

Are you sure the cooler is working as intended. Pump working? Both pipes roughly same temp? One colder than the other (to touch)?
 

biL4

Reputable
Feb 27, 2019
19
1
4,515
But you're clearing CMOS after each bios update? Yes?

Are you sure the cooler is working as intended. Pump working? Both pipes roughly same temp? One colder than the other (to touch)?
just cleared CMOS again

left cinebench running for 15 minutes and both pipes have the same temp. I suppose the pump is working otherwise it would be much worser
 

Karadjgne

Titan
Ambassador
Depends on the Prime95. Small fft uses extremely little other than the cpu, as you go larger, more is used, like ram, so you'll not get a good 100% load. Prime95 by default also uses AVX technologies, you must uncheck that box/boxes and disable AVX. As is, AVX is an instruction set for vector analysis, so is extremely cpu intensive, offering a 'virtual' upto 130ish% cpu load. It's brutal on a cpu and the resultant temps, which is why it should be disabled as it's unrealistic to 'normal' usage. Most games don't use any AVX, a few do in combination with PhysX, for the particals flying in every direction after an explosion, because it's a faster process than is normal.

So if running Prime95, it should be small fft AVX disabled, any other run won't give reliable or realistic values.
 

biL4

Reputable
Feb 27, 2019
19
1
4,515
Depends on the Prime95. Small fft uses extremely little other than the cpu, as you go larger, more is used, like ram, so you'll not get a good 100% load. Prime95 by default also uses AVX technologies, you must uncheck that box/boxes and disable AVX. As is, AVX is an instruction set for vector analysis, so is extremely cpu intensive, offering a 'virtual' upto 130ish% cpu load. It's brutal on a cpu and the resultant temps, which is why it should be disabled as it's unrealistic to 'normal' usage. Most games don't use any AVX, a few do in combination with PhysX, for the particals flying in every direction after an explosion, because it's a faster process than is normal.

So if running Prime95, it should be small fft AVX disabled, any other run won't give reliable or realistic values.
Thanks for the information on Prime95. Currently I'm using just Cinebench R23 because its enough to get the CPU over 85C. If I can figure out why this is happening and fix it probably I will switch to Prime95 for a stress test and using those settings
 

biL4

Reputable
Feb 27, 2019
19
1
4,515
ended up with this PBO2 Tuner

-30 Curve on All Cores

Limits
PPT 114W
TDC 75A
EDC 115A

Sustained 4.4GHz during Cinebench R23 and max. temps of 80C
 
Last edited: