Announcement

Collapse
No announcement yet.

Linux 5.10-rc2 Released Following The Intel MIC Removal

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Linux 5.10-rc2 Released Following The Intel MIC Removal

    Phoronix: Linux 5.10-rc2 Released Following The Intel MIC Removal

    The second weekly release candidate to Linux 5.10 is now available for testing...

    http://www.phoronix.com/scan.php?pag...0-rc2-Released

  • #2
    Greg Kroah-Hartman drops the MIC with the release of Linux 5.10-rc2

    Comment


    • #3
      I tried rc1 and rc2 (Ubuntu 20.10) and both of them drive up the Vcore on my Ryzen 3900X to 1550 mV in idle

      Comment


      • #4
        Originally posted by mumar1 View Post
        I tried rc1 and rc2 (Ubuntu 20.10) and both of them drive up the Vcore on my Ryzen 3900X to 1550 mV in idle
        That's an excellent catch. Here's a quick screenshot from some light vid/almost idle behaviour:



        5.9 on the left, 5.10-rc2 on the right. 3700x But now that you mention it, it almost looks like Vsoc and Vcore are being mixed up AND misreported?
        Is something actually causing increasing power draw or is it just misreporting? I'll do some git regression testing later this week if it's still a problem - there's patches to k10temp quite recently that might've done it.
        https://git.kernel.org/pub/scm/linux...wmon/k10temp.c

        Since I don't think the chip would allow (Over voltage protection surely must come into play) 1.55V on it, especially 'idle', it's more likely just misreporting. Still, might avoid 5.10 til this is looked into further.
        Last edited by Linux_Chemist; 02 November 2020, 08:55 PM.

        Comment


        • #5
          Originally posted by mumar1 View Post
          I tried rc1 and rc2 (Ubuntu 20.10) and both of them drive up the Vcore on my Ryzen 3900X to 1550 mV in idle
          Thank you again for highlighting this, I've narrowed it to commit d6144a40041a70ecce410a07d3a2b6906f6e7da9
          Bug reported is here if you'd like to keep an eye on it
          https://bugzilla.kernel.org/show_bug.cgi?id=210121

          Until it's confirmed whether it's just misreporting the voltage and not actually an accurate output (of a voltage way too high to have permanently at idle), best to avoid latest 5.10-rc kernels for zen 2 cpus imo. Or revert the commit of course

          Comment

          Working...
          X