Wildcard matching on temperature source identifier

[kevinstampe]

Sometimes my gpu core temp gets recognized with identifier /gpu-amd/0/temperature/0, sometimes with /gpu-amd/1/temperature/0.

I would like to either get a wildcard matching, with either a * or even regex matching would be fine. (example /gpu-amd/*/temperature/0)

Relevant hardware specs and setup
ASUS Strix x570-e mobo, Ryzen 5800, 64gb ram, Lenovo Legion RX6800xt gpu

[Rem0o]

Ugh, the problem is that I can't do that, it won't work on multi-card setup. The number that is changing is called the adapter index. Never saw it moving like that. Do you physically swap the card from a slot to an other? Did you change drivers? Anything that might ring a bell? Do you have multiple gpus?

[kevinstampe]

I only have a singular gpu, no physical swapping or anything. I have 3 pci-e devices, where only one is gpu. The other two is m.2 nvme ssds.
It happens frequently after updating amd drivers.
Do you have any obvious way of debugging this behaviour? Im a software developer, so im not afraid to dig into this.

You could just say .FirstOrDefault() on devices, this would only be for edge case handling though. So no UI support, only for manually editing in config file.

[Rem0o]

Load up LibreHardwareMonitor, the part that look at adapter index is in here:
https://github.com/LibreHardwareMonitor/LibreHardwareMonitor/blob/master/LibreHardwareMonitorLib/Hardware/Gpu/AmdGpuGroup.cs

[kevinstampe]

I've written some sample code to test this, though i cannot test this right now. ATM my gpu is index 0, and no other hardware is found that could bump this index.
When FanControl yells at me again (happens atleast once a month), i will test this, and see what is in index 0.
I'm pretty sure it happens when i update AMD Adrenalin, though can't reproduce this by downgrading and then upgrade again.

Can we keep issue open, until my debugging is done? Thanks

[Rem0o]

^ One of my guess is that when you upgrade adrenalin, it allocates the new driver as a new gpu, then disable the "old" one, which creates a new index until you reboot.

[kevinstampe]

This actually happens after reboot, when i installed new driver. In one of your earlier versions, it still displayed the graphs, just with empty temp sensor.
Now it removes the graph, which makes this problem worse for me.

And the error dialogue tells me that there are no data to fetch from /gpu-amd/0/temperature/0.

This tells me that there are no device on index 0 for some reason.

But there is data present on /gpu-amd/1/temperature/0.

I still think that this could be fixed with manual explicit pattern matching in the .json file.

[Rem0o]

^ I can't pattern match, this would break a whole lot of other stuff. Identifiers are meant to be constant, and unique, assuming fixed hardware.

[kevinstampe]

And i get that, but identifiers can also be searchable in a wider context, than just static written in a file.
And when the software for the hardware shits it's bed and creates another identifier, thats not on you. It's either on AMD or on LibreHardwareMonitor - but this could be fixed with an .SingleOrDefault(). Then if null, throw usual error - if more than one is found, show another error.
Do you perhaps have the source repo public? Would love to fork it, and show what my idea is.