New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
CPU Temps not showing for AMD Ryzen Threadripper 3970X. #1484
Comments
@iliarostovtsev can you take a look at this? It looks like you most recently updated the code to detect CPU temps on Ryzen. |
Fixed - 7fe97a7 |
It looks that you're having encoding issues in your shell. |
Sorry about the munged characters. Looks like that happens if you use putty through RDP. I've attached a corrected version just in-case. Thanks for looking into it. |
The patch above will work. Thanks. |
Using the latest version of webmin didn't fix this issue:
Here's the "sensors >sens.bin" from my system saved and transferred as binary. Note: the github webUI won't let me upload a *.bin file, so I've renamed it to sens_bin.txt. Looking at the HEX view of the file, the degree symbol is encoded using 2 bytes, 0xC2 and 0xB0. I tried to manually grep sensors using the new line from /proc/linux-lib.pl and it seems to work.
So I'm not sure what's wrong. |
What is the full output of |
The attached sens_bin.txt file in the message above is the full output. I'll quote a copy here also and hope characters don't get corrupted.
Thanks! |
Thanks for reporting. Give this patch a try, please - |
No, it still doesn't work. Looking at that section of code a bit more, I don't think it's the degree character.
The thread ripper sensors output doesn't include any fan or voltage data in the output, so the "if ($aa && " in the temp parsing is failing before it ever gets to parse out the degree symbol. I just tried running sensors-detect again and running scans for every device I could, it still shows the exact same output as the attached file, with just one CPU temp. I know there are a bunch of fans inside that system too, which is a little sad that I'm not able to get info for those either. |
.. so, this part:
is completely unrelated to CPU? |
Does your sensor command provide CPU temps at all? |
The k10temp-pci-00c3 I believe is the CPU temp sensor. There are other sensors on the motherboard, temp, fan and voltage that would normally be displayed, but are apparently not currently supported for the Super IO chip used by my ThreadRipper motherboard. Here's the output from the sensors-detect that I believe shows the missing kernel driver that I need:
I believe the nct6775 kernel driver may have some support for my system, so I'm currently trying to figure out how to get it to load. |
Well, I was able to get my fan speed and voltages to show up in the sensors output, but it still won't show a temperature in webmin. :(
|
Which line(s) exactly on the output above represent CPU temperature? |
The k10temp-pci-00c3:temp1 temperature is the main temp reported by the CPU. It doesn't require installing extra drivers. However it apparently has reliability issues: https://wiki.archlinux.org/title/lm_sensors#Troubleshooting The nct6797-isa-0a20 sensors only showed up on my system when I activated https://elrepo.org/ and installed the nct6775 driver from that repo. These are sensors that are provided by the motherboard and I'm not exactly sure how they relate to the actual CPU temperature. Just from my own evaluation, I can see the number that most closely matches the k10temp is the SMBUSMASTER_0 temperature. The SYSTIN temp say's it's a CPU diode, and seems reasonable as a cpu temp, but is much lower than the k10temp. The CPUTIN temp seems very low and almost never changes, so I'm not sure what that's measuring. I wish I had a better answer for you as to which temp to read. I think the k10temp sensors is the best one, since it doesn't require extra drivers and would give the broadest support. |
By modifying the code to ignore the fan and voltage requirement I was able to get the temperature to show up in webmin.
|
This the the temperature read from the CPU sensor on the MB. We need CoreTemp data. This sensors command output is just odd in my opinion. Check this patch - 76591fa - replace the whole file to avoid typos. |
Check this patch out - d10e122 |
Ok, says Core 1 now. Thanks! |
Cool server! |
Because of all the non-standardized names/numbers of temp sensors, many temp displaying apps let the user pick which temp sensor to display, in case the app logic guesses "incorrectly"... For an app as an example, on the PC/Windows, the app "Speedfan"... |
I was thinking of adding it but it would require an extra configuration option but we already have so many. |
The text was updated successfully, but these errors were encountered: