No announcement yet.

Monitoring RX Vega temperature in Linux

  • Filter
  • Time
  • Show
Clear All
new posts

  • Monitoring RX Vega temperature in Linux


    Recently got an RX Vega 64. Will be water cooling it in the near future.

    I would like to see the temperatures - more than one - as I suspect it might be usefull to spot any installation errors after I've installed the water cooler.

    I'm under the impression that in Windows the AMD GPU driver control panel gives more information, as does GPU-Z in Windows:

    However, current drivers seem to only show one temperature:

    Adapter: PCI adapter
    vddgfx:       +1.09 V  
    fan1:        2393 RPM
    temp1:        +81.0°C  (crit = +91.0°C, hyst = -273.1°C)
    power1:      194.00 W  (cap = 220.00 W)

    Only relevant discussion I've been able to find is this:

    But it seems that hotspot temperature patch has not propagated to the current Kernel (4.19.4) ?

    Am I being overly suspicious? Is it enough to monitor just one temperature (and be done with it) ?

  • #2
    Debianxfce: are you saying they have added sensor readings in more recent / testing branch Kernels?

    From the looks of it, there are no changes to incorporate any more sensors (besides temp1) here:

    Nor can I find anything in the changelog, but I could be looking at the wrong changelog.

    I find it a bit irresponsible to randomly suggest something quite labor intensive unless you know otherwise, and more senor data has actually been implemented on the bleeding edge.

    Bottom line: I'd rather not compile my Kernel "just to try it out"; that's why I'm asking here. How is it on your system, how much sensor data do you get?

    FWIW, I forgot to mention, this is on Arch Linux. 4.19.8 is in testing; I could temporarily install that Kernel, and of course I could always recompile the kernel or the module (with the patches linked in my first post), I was just wondering if there is another way, or ways to get even more data besides the hotspot...


    • #3
      Originally posted by debianxfce View Post
      That does not solve the problem. That's only a workaround.


      • #4
        tildearrow : agreed. However, an IR thermometer might come in handy in any case, since the card or other HW might not have sensors at all interesting locations. Also, it can come in handy in other activities possibly having nothing to do with graphics cards or PCs ;-). Found one for cheap at a local HW store.

        An IR thermometer is actually useless in this case, since important parts are covered by the cooler, hence impossible to measure.

        Still looking for other solutions (save applying the patch and recompiling the modules - and to get more than two sensors of data, which these cards should gave, seems impossible ATM).