I have a multiboot system. One of the installed OS’s does not use the NVMe SSD installed on the motherboard at all.
At the time of taking the screenshot, all the SSD partitions are unmounted, so apart from detection, the SSD is mostly unused.

  • I would like the temps to drop down to SYSTIN (≈35°C) levels.
  • I know, it’s right next to my GPU, but I am not doing anything GPU intensive, the GPU temps are ~37°C [1]

For the unmounted and unused HDDs, I just use hdparm -Y, but there seems to be nothing in terms of that for the SSD. And even though I appreciate the additional heat in winters, this is going to be too expensive for me. I’d rather burn some cheap Nichrome than my data storage device.

I checked out a Debian forum thread and from that, I checked the following:

❯ sudo nvme get-feature /dev/nvme0 -f 2 -H
get-feature:0x02 (Power Management), Current value:0x00000004
        Workload Hint (WH): 0 - No Workload
        Power State   (PS): 4

Showing it is already in the lowest power state.

I have no active cooling setup for the SSD from my side. This becomes relevant soon.

  • Checking the SSD temps (using the same widget as in the image), the temperature on Sensor 2 starts out at ~40°C (after a normal reboot) and slowly increases to >50°C as shown at the start of the graph. Power State (PS) is still 4.

  • Running KDE partitionmanager, which probably does some reading to check the partition information, at 50°C stage, causes a temperature drop, as shown in the image.

  • Running KDE partitionmanager right after reboot, when the temperature is increasing very sloowly, seems to do nothing significant.

  • Turns out that after a few minutes of System Standby, the SSD doesn’t return to PS: 4, so I have the culprit.
  • Running partitionmanager after that causes it to go back to PS: 4

So we have a solution! All I need to do is run partitionmanager on wake. nlol jk

Motherboard: MSI MAG X570S TORPEDO MAX (MS-7D54)
SSD: Samsung 980 512GB (correct firmware, bought long before the fakes started coming out)

  1. apart from GPU memory, which is 48°C due to the awful AMD 7th gen Zero RPM, which has no workarounds on Linux ↩︎

  • nanook@friendica.eskimo.com
    3·
    18 days ago

    I’ve got a Gigabyte X299 Aorus Master with i9-10980xe overclocked at 4.8Ghz, two nvme drives mounted on the motherboard, this is about the most power hungry consumer level CPU you can get (maxes at around 540 watts), yet the ssd’s which are also fairly heavily used in a database, only run at about 29C. Perhaps your GPU is heating up your SSDs? I have no active cooling on mine either, just the cheesy heat sinks. I’m running 6.11.2, unfortunately 6.11 kernels past two have some code that breaks the i9-10980xe so stuck at that for the moment.

    • ulterno@programming.devOPEnglish
      1·
      17 days ago

      Perhaps your GPU is heating up your SSDs?

      Of course, that is happening to an extent, but the moon is not hotter than the sun, right?

      Also, I later found out, the problem was that the SSD was staying on when I woke from Sleep.

      As to why my OFF temp is higher than your ON temp, it’s probably just bad airflow, which I haven’t come to fix yet.

    • ulterno@programming.devOPEnglish
      1·
      17 days ago

      Finally, yaay!
      Thanks to Wolfgang Müller.

      My GPU Zero RPM stayed on until 60°C and considering I lost my laptop to heat just last year, this was pinching me now and then.

    • ulterno@programming.devOPEnglish
      0·
      15 days ago

      Of course a drive heats up with use.

      As I explained in the wall of text above, it was heating when NOT in use.
      The reason being that after wake from S3 Sleep, it was turning on and not turning back off (as it should because it was NOT MOUNTED).

      I just need to check which process is responsible for the SSD’s power state management (systemd maybe?) and then see if something can be changed.