Question Just installed a "Ryzen 5800X", are these temps normal?

Mr Drummer

Distinguished
Sep 21, 2015
134
2
18,685
I just installed a "Ryzen 5800X" on my "MSI B550-A PRO" motherboard. The CPU is cooled by a "Noctua NH-D15" with only one fan (in the middle). I upgraded from a "Ryzen 3700X" which ran very stable and cool (27-29C in idle) and i'm just wondering if someone can have a quick look at temps of my new 5800X. The first picture shows the temps i get when i first start HWinfo64, the second shows the temps after about a minute with just HWinfo64 and NordVPN open, the third the temps after watching a 4K video on Youtube for about 2-min. Are these temps normal or unusually high? I've done no overclocking, only set an XMP-profile in the BIOS and the energy setting set to "AMD Ryzen Balanced".


1.

2.

3.

Thanks in advance!
 
Last edited:
So, a few things here.

One, just because "you" aren't overclocking, doesn't mean your system isn't overclocked unless you've disabled PBO. PBO, IS "overclocking", it's just automatic overclocking done by the system based on very specific criteria determined by the engineers that designed the hardware. And with a D15 there is absolutely no reason you shouldn't be able to leave PBO enabled without any worries about cooling problems.

Which leads us into number two. And this is kind of a two-parter. Why are you running only a single fan on that D15? Or is it actually a D15s? And for just watching video, whether 4k or any other resolution, 70°C on the CCD1 (Tdie) seems pretty high. But you have to also factor in whether there were other processes, potentially using a fairly high load, running at the same time in the background. If Windows was running an automated task like disk optimization, creating a system restore point or running Windows update, at the same time you were watching that video, it MIGHT account for that.

The very first thing you should do if you have not already is make sure you are on the latest BIOS version 7C56vAF from 10-12-2023 installed.

The second thing you should do is make sure you have the latest chipset driver installed from the AMD website.

https://www.amd.com/en/support/chipsets/amd-socket-am4/b550


Also, it would be additionally helpful to know exactly how many case fans are installed, exactly WHERE you have them installed, what model they are and in exactly what orientation, intake (in) or exhaust (out) each of them is configured for?
 
  • Like
Reactions: CountMike
I just installed a "Ryzen 5800X" on my "MSI B550-A PRO" motherboard. The CPU is cooled by a "Noctua NH-D15" with only one fan (in the middle). I upgraded from a "Ryzen 3700X" which ran very stable and cool (27-29C in idle) and i'm just wondering if someone can have a quick look at temps of my new 5800X. The first picture shows the temps i get when i first start HWinfo64, the second shows the temps after about a minute with just HWinfo64 and NordVPN open, the third the temps after watching a 4K video on Youtube for about 2-min. Are these temps normal or unusually high? I've done no overclocking, only set an XMP-profile in the BIOS and the energy setting set to "AMD Ryzen Balanced".


1.

2.

3.

Thanks in advance!
Don't see anything wrong with those temps except at "Idle" but you are not showing frequencies or how really is it Idle. That should be no more than 1 eventually 2% of CPU load. It's a way hotter CPU than 3700x, allowed up to 90-95c to retain full performance vs 3700x at 60-65c. Only time to worry is if temps approach or reach 95c. Under that temp it doesn't matter at all.
 
Last edited:
Don't see anything wrong with those temps except at "Idle" but you are not showing frequencies or how really is it Idle. That should be no more than 1 eventually 2% of CPU load. It's a way hotter CPU than 3700x, allowed up to 90-95c to retain full performance vs 3700x at 60-65c. Only time to worry is if temps approach or reach 95c. Under that temp it doesn't matter at all.
Ok, good to know. I took two snapshots just now of most of the Hwinfo64 (the second picture showing the frequencies i believe). I hear the CPU-fan ramp up a bit sometimes (still very quiet though) and it's not always when i do something demanding.



 
Don't see anything wrong with those temps except at "Idle" but you are not showing frequencies or how really is it Idle. That should be no more than 1 eventually 2% of CPU load. It's a way hotter CPU than 3700x, allowed up to 90-95c to retain full performance vs 3700x at 60-65c. Only time to worry is if temps approach or reach 95c. Under that temp it doesn't matter at all.
what people see as idle depends on them, not on some specific number. Not all PC idle at 2%, it can depend on what applications are running. I generally idle around 3 to 5%.

HWINFO itself adds to the idle state of the PC, its sensors do. Its most of my idle
v2AeYbT.jpg


Ryzen CPU will wake a core to do a function, run it as fast as possible to complete task, and then rush back to idle again, to save power... So they constantly bouncing around in temps.

Your temps look like mine... 5800x3d
zZIYkt4.jpg
 
Last edited:
what people see as idle depends on them, not on some specific number. Not all PC idle at 2%, it can depend on what applications are running. I generally idle around 3 to 5%.

HWINFO itself adds to the idle state of the PC, its sensors do. Its most of my idle
v2AeYbT.jpg


Ryzen CPU will wake a core to do a function, run it as fast as possible to complete task, and then rush back to idle again, to save power... So they constantly bouncing around in temps.

Your temps look like mine... 5800x3d
zZIYkt4.jpg
Idle is running just what's minimally necessary to operate at all, just OS and drivers,anything else will wake up cores or drive used cores harder. Even test programs do that.
 
Your temps aren't that bad now.

I average 33c at idle on cores, package is 40c (its the CPU sensor that shows in Motherboard area). I can get to 29c on cores if I let my AIO fans run on their default plan but I don't need that low.

You can't avoid heat completely, I find CPU hits 70c at boot... but thats only time its up there. Windows is loading everything all at once, its going to use all the CPU.
 
  • Like
Reactions: CountMike
Idle temps do not, and have NEVER mattered, on any generation or platform of CPU, EXCEPT in cases where the maximum thermal specification is being exceeded. It would not matter if your CPU was idling at 70°C if TJmax is 85°C and you never exceeded that no matter what kind of load you put on it. Idle temps, don't matter. They DO indicate a problem though if you are exceeding maximum thermal specification for your CPU model.

But that also doesn't mean you can have some kind of problem even if you are not exceeding maximum thermal spec but are seeing higher than normal temps under specific conditions that should not result in those kind of temps, such as a semi-bad paste job or CPU cooler mount, or internal degradation or other issues with the CPU itself. Sometimes just firmware can cause problems if the spec voltage has been set incorrectly.

Also, it would be helpful in your HWinfo screenshots if you would click the arrow next to core clocks and core temps so we can see on an individual core basis what is going on, because sometimes it's simply a bad core causing problems and you can't see that without seeing what each core is doing. Ten degrees or more variance between cores is generally a warrantable condtion for both Intel and AMD CPUs.

IyCuUXH.png
 
  • Like
Reactions: CountMike
Idle temps do not, and have NEVER mattered, on any generation or platform of CPU, EXCEPT in cases where the maximum thermal specification is being exceeded. It would not matter if your CPU was idling at 70°C if TJmax is 85°C and you never exceeded that no matter what kind of load you put on it. Idle temps, don't matter. They DO indicate a problem though if you are exceeding maximum thermal specification for your CPU model.

But that also doesn't mean you can have some kind of problem even if you are not exceeding maximum thermal spec but are seeing higher than normal temps under specific conditions that should not result in those kind of temps, such as a semi-bad paste job or CPU cooler mount, or internal degradation or other issues with the CPU itself. Sometimes just firmware can cause problems if the spec voltage has been set incorrectly.

Also, it would be helpful in your HWinfo screenshots if you would click the arrow next to core clocks and core temps so we can see on an individual core basis what is going on, because sometimes it's simply a bad core causing problems and you can't see that without seeing what each core is doing. Ten degrees or more variance between cores is generally a warrantable condtion for both Intel and AMD CPUs.

IyCuUXH.png
Hi, here i've expanded the view in HWinfo64 of what i think you're referring to. I don't have something called "Core temps" though but maybe i have to check something in the settings for it to show up or you have another version of the program (i don't know). But here is what i can expand upon so to speak, and these are pictures showing my PC in it's "idle state".

1.
2.
 
Ok, but "idle state" doesn't tell us anything really. What matters is steady state, full, all core load.

No different than if you were overclocking, even if you are not and everything is at the stock configuration. It's still the same thermal compliance testing.

Looks like I need to fix some image hosting too.


 
  • Like
Reactions: CountMike
Ok, but "idle state" doesn't tell us anything really. What matters is steady state, full, all core load.

No different than if you were overclocking, even if you are not and everything is at the stock configuration. It's still the same thermal compliance testing.

Looks like I need to fix some image hosting too.


Here are two pics showing HWinfo64 after about 10-mins of stresstesting (OCCT, the extreme option). To note is that i've also done a light undervolting of the "Curve Optimizer" in the BIOS here (-20 on all cores). Something i'll maybe fine tune abit as time goes on. I also ran a full scan of Windows Defender, my brother suggested that this puts the most load he's seen on a CPU, and it reached a maxtemp of 80c.

1.

2.
 
Windows defender actually has very little actual core utilization. It might run one or two cores up to about 50% utilization but more than anything it relies primarily on storage device reads. So, no, it is not a good option for thermal or stability testing, which are in fact entirely different things even though both can often be done simultaneously. The key is steady state and there are only a few utilities that actually impose an all core steady state load. OCCT CPU small data set, Prime95 Small FFT (Not "Smallest FFT"), Heavyload stress CPU, Furmark CPU burner, Intel processor diagnostic tool, Aida64 system stability test stress CPU.

Although, as I was informed by my friend and author of the Intel temperature guide Computronix, the Aida64 stress CPU test is basically worthless in terms of stressing the CPU for thermal testing as it fails to come anywhere near TDP so it's not a good choice for that. Which, basically means I ignore it and recommend others to do the same since there are a number of others that don't have that issue. The last thing you want is a false sense of security because a utility is not actually doing what you think it is doing AND in regard to that and going back to the Windows defender consideration, keep in mind that core frequency can all be maxed out without any actual utilization of that core happening. Just because the core is running at it's max boosted frequency does not mean the core is actually doing any work and if it's not doing any work then it's useless as a metric for testing stability or thermal compliance.

Also, if the work is not "steady state", meaning it does not fluctuate up and down like many loads do, then it is also useless for thermal compliance testing as the cores get too much chance to "rest" when it is not a steady state load. You can test stability with changing loads, but thermal testing should always be done with a steady state load that does not change from 100% until you tell it to. Generally about ten minutes of full load will reach whatever saturation it's going to by then.

OCCT extreme option MUST I believe be configured for "small data set" otherwise you are not running the correct configuration for thermal testing. It's also a good idea to set OCCT's settings for critical temperature to like 85-90 degrees depending on what CPU is being tested as well as "stop on first error found" as there is no point in continuing to run any test once an error has occurred. Some shouldn't really be run past 80-85°C while some of the newer CPUs should be fine up to about 95-100°C although personally I wouldn't want to see anything over 95°C even on newer CPUs. It's just pushing the envelope too far in my opinion and getting into throttle territory.

Your temps look ok running OCCT, although I don't know if you were running the correct tests or not, but there is no way you should have been seeing 80°C temps running Windows defender, ever, unless you were also running other high load processes at the same time.
 
  • Like
Reactions: CountMike
Windows defender actually has very little actual core utilization. It might run one or two cores up to about 50% utilization but more than anything it relies primarily on storage device reads. So, no, it is not a good option for thermal or stability testing, which are in fact entirely different things even though both can often be done simultaneously. The key is steady state and there are only a few utilities that actually impose an all core steady state load. OCCT CPU small data set, Prime95 Small FFT (Not "Smallest FFT"), Heavyload stress CPU, Furmark CPU burner, Intel processor diagnostic tool, Aida64 system stability test stress CPU.

Although, as I was informed by my friend and author of the Intel temperature guide Computronix, the Aida64 stress CPU test is basically worthless in terms of stressing the CPU for thermal testing as it fails to come anywhere near TDP so it's not a good choice for that. Which, basically means I ignore it and recommend others to do the same since there are a number of others that don't have that issue. The last thing you want is a false sense of security because a utility is not actually doing what you think it is doing AND in regard to that and going back to the Windows defender consideration, keep in mind that core frequency can all be maxed out without any actual utilization of that core happening. Just because the core is running at it's max boosted frequency does not mean the core is actually doing any work and if it's not doing any work then it's useless as a metric for testing stability or thermal compliance.

Also, if the work is not "steady state", meaning it does not fluctuate up and down like many loads do, then it is also useless for thermal compliance testing as the cores get too much chance to "rest" when it is not a steady state load. You can test stability with changing loads, but thermal testing should always be done with a steady state load that does not change from 100% until you tell it to. Generally about ten minutes of full load will reach whatever saturation it's going to by then.

OCCT extreme option MUST I believe be configured for "small data set" otherwise you are not running the correct configuration for thermal testing. It's also a good idea to set OCCT's settings for critical temperature to like 85-90 degrees depending on what CPU is being tested as well as "stop on first error found" as there is no point in continuing to run any test once an error has occurred. Some shouldn't really be run past 80-85°C while some of the newer CPUs should be fine up to about 95-100°C although personally I wouldn't want to see anything over 95°C even on newer CPUs. It's just pushing the envelope too far in my opinion and getting into throttle territory.

Your temps look ok running OCCT, although I don't know if you were running the correct tests or not, but there is no way you should have been seeing 80°C temps running Windows defender, ever, unless you were also running other high load processes at the same time.

I set OCCT extreme to "small data set" (as you suggested) today after work just for fun and this is what i got.





 
Ok, but per your screenshot you had it set to "variable" rather than "steady". As I said, "steady state" IS the key to accurate thermal compliance testing. I would try it again AND also make SURE you have SMALL data set, STEADY state and check through the OCCT settings to make sure you have DISABLED ALL AVX types including AVX, AVX2 and AVX512 before running the test.
 
Ok, but per your screenshot you had it set to "variable" rather than "steady". As I said, "steady state" IS the key to accurate thermal compliance testing. I would try it again AND also make SURE you have SMALL data set, STEADY state and check through the OCCT settings to make sure you have DISABLED ALL AVX types including AVX, AVX2 and AVX512 before running the test.
Yeah, you're right i missed that, thanks for pointing it out.
 
  • Like
Reactions: Darkbreeze