Announcement

Collapse
No announcement yet.

Some Ryzen Linux Users Are Facing Issues With Heavy Compilation Loads

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

  • #21
    It shouldn't be too hard to figure out if it's an unstable overclock. Run mprime and if it survives that for a day it's not your overclock settings.
    My Bulldozer build (on a Crosshair V) was not mprime-stable at stock; it needed a little voltage nudge.

    Comment


    • #22
      I tend to agree with those that say it is probably unstable overclocking. My 1800x is perfectly stable and can rebuild the entire @world on Gentoo without issue, but I run it at 3800MHz maximum on all cores. It isn't reliably stable above that in Gentoo or Windows (which my kid uses to play games). Also, I'm only running 2667MHz ram, since that is the maximum the board said it supported when I bought it.
      Overall, I don't think this platform is very good for overclocking, but is plenty powerful without it.

      Comment


      • #23
        I don't think all these people are overclocking. I'm certainly not, unless you count enabling XMP to 2667MHz, which isn't that crazy. Far more likely that many of them are using -march=native or -march=znver1.

        Comment


        • #24
          That doesn't make any sense, if there's an issue for some users, it could be anything, but not the CPU, as other said it could be BIOS, could be unstable CPU/RAM overclock, could be GCC itself.

          Comment


          • #25
            Originally posted by Alliancemd View Post
            (we are software developers) ... That would of been a bad decision...
            but would it have been as bad of a decision as hiring a software developer who says "would of"

            Comment


            • #26
              Originally posted by rrohbeck View Post
              It shouldn't be too hard to figure out if it's an unstable overclock. Run mprime and if it survives that for a day it's not your overclock settings.
              My Bulldozer build (on a Crosshair V) was not mprime-stable at stock; it needed a little voltage nudge.
              I run my bulldozer massively undervolted at 1.165v with stock clocks, never had a problem in compilation. As soon as i go down to 1.143v i get those problems.

              Comment


              • #27
                Not having this issue running a 1700x @ 3.9. Like others have stated, Unstable OC by the sounds of it. OBS with CPU encoding is good at showing a unstable OC too

                Comment


                • #28
                  OK, maybe I'm being naive here, but if someone was running a system with overclock and having any kind of stability problems I figured the first thing they would do is run at stock clocks for a while to ensure the problem still happened before thinking there was a problem...

                  ...even typing that I feel like someone is snickering

                  My initial impression is that most (but not all) of the people seeing problems are running Gentoo - is that the impression that others are getting as well ?
                  Last edited by bridgman; 06-02-2017, 07:39 PM.

                  Comment


                  • #29
                    I've been having issues with compilation of Mesa segfaulting on me occasionally as well.. usually I can just start make again, and it runs a to completion. I am running a ryzen 1700, no cpu overclock, can't remember the exact memory timings... If it's overclocked memory at all, it's 2667/2933mhz with the first XMP profile.


                    ​​​​​​I do have GCC 6.3, and I am using March=native... I'll maybe try GCC vs clang tests after the kid's in bed to see if it's a GCC issue.

                    Edit: Seems I'm unable to reproduce anymore, but I have done a BIOS update and reset stock settings there in the last week or two, so it's possible that the issue was related to the memory clocks/timings I had selected... If it happens again, I'll try to figure it out.

                    All that being said, since the first few BIOS updates came out, my system has been pretty stable, and hasn't had any major issues. The GCC thing I figured was due to a threading/race issue in the compiler, since I could just switch to clang instead and my builds would complete reliably. It wouldn't be the first time that GCC 6.3.0 screwed things up for me.
                    Last edited by Veerappan; 06-03-2017, 04:47 PM.

                    Comment


                    • #30
                      Originally posted by Chewi View Post
                      I don't think all these people are overclocking. I'm certainly not, unless you count enabling XMP to 2667MHz, which isn't that crazy. Far more likely that many of them are using -march=native or -march=znver1.
                      FWIW, I'm also using -march=znver1.

                      Pretty sure that these people are seeing problems with insufficient voltage, since some note that a change to LLC fixes the problem - and it wouldn't surprise me even a little bit if some of the BIOSes out there are don't have the voltages right just yet.

                      Comment

                      Working...
                      X