CPU [#0] Bug Threadripper 3000 and 7000

[Warcobra]

Hi,

I noticed that no matter if I use a 3960x or a 7960x, HWINFO64 will eventually cause me to lose visibility of the CPU [#0] sensors and, on top of that, will cause my CPU to no longer turbo. The only way to fix the turbo issue it to restart the machine. I honestly thought it was just my older Threadripper, but then I tested this on multiple different machines I have access to, and it will happen eventually. I have a debug file from when it's happening from my 7960x and could get one from my other threadripper, but it's the same issue. I am also attaching screenshots of the portions of HWINFO64 sensors that I lost. These images are from before it happened, to be precise. If anyone has an idea as to why this happens, I would be happy to test things and eventually figure out why. I use HWINFO64 to monitor many things and would like to continue using it, but it's complicated if it's going to cause an issue with my CPU turbo states. I have also been running HWINFO64 on Ryzen and Intel machines alike with no issues running many of the same services on each machine, and they are all fine. So, this seems to be a bug for Threadripper within the last couple of months, as this has been happening for about that long for me.

Thanks in advance

PS: I had to compress the debug file because it was too large.

 

[Martin]

Are you perhaps running some other monitoring/tweaking tool along with HWiNFO?
According to the dump you provided it seems that some other software is blocking HWiNFO. Moreover, it seems to heavily accesses some CPU interfaces, so HWiNFO can't access them.

 

[Warcobra]

So, the only thing I have running on the machine from which this dump is from is a Stream Deck. It has no active monitoring functions installed. It's just for hotkeys for editing pourposes. The issue is that when I have the sensor bug of some sort, all monitoring software can no longer access CPU sensors. So CPUZ etc, can no longer see anything thing from the CPU until I restart my PC. Where else would you suggest I look? I am considering installing a fresh Windows copy on another machine again and seeing if it happens. The only reason I think it's not is that it happens on four other machines that are running Threadripper. Everything that's not Threadripper works fine.

 

[Warcobra]

Here is a debug right after restarting before having the issue for comparison.

 

[Martin]

Not even ASUS AI Suite is running?
So the problem appears only after certain time?

 

[Warcobra]

Correct, I am not running anything else that would be reading sensor information. Also yes, it can happen anywhere between 1 to 5 days anywhere in between. I have yet to make it past 5 days without it happening. Of course, if I don't run it at all, I'm fine. But to be fair, I have yet to run any other senior monitor long enough to see if it happens.

 

[Martin]

Well, this could also be due to some problem in the SMU of the CPU as it seems to stop responding. Perhaps upgrading the BIOS could help.
But I saw there some other strange things that could indicate a problem with other monitoring software that blocks HWiNFO. So it would be interesting to run some other tool (e.g. HWMonitor) and see if the same will happen.

 

[Warcobra]

Im going to start running some other software to monitor to see if it causes the same thing. I also found some other posts in other forms about people having the issue where their SMU locks up and stops reporting, along with freezing the clocks at low speeds. This is what's essentially happening to me. So, I will report back when I have some new findings. Thanks for the assistance so far Martin.