Date: Sun, 08 Oct 2017 10:44:42 +0000 From: bugzilla-noreply@freebsd.org To: freebsd-bugs@FreeBSD.org Subject: [Bug 219399] System panics after several hours of 14-threads-compilation orgies using poudriere on AMD Ryzen... Message-ID: <bug-219399-8-wis3HAg1pb@https.bugs.freebsd.org/bugzilla/> In-Reply-To: <bug-219399-8@https.bugs.freebsd.org/bugzilla/> References: <bug-219399-8@https.bugs.freebsd.org/bugzilla/>
next in thread | previous in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D219399 --- Comment #255 from Mark Millard <markmi@dsl-only.net> --- (In reply to SF from comment #254) Just reporting a distinct example. . . The Ryzen 1800X system that I had access to for a time had water-based cooling that kept it under 47 degC even during hot summer days with no air conditioning but doing builds. (Clearly not true of any days getting near that temperature in the first place. But it gives a clue as to effectiveness.) The power supply was also good. There was only a basic, low power video card present. That did not stop the SIGSEGV's, missing temporary files, having lang/ghc almost always fail to build, etc. Although, in general, I seemed to get them somewhat less often than some folks seemed to be reporting --other than lang/ghc nearly always failing fairly rapidly. [Side question: Have you tested your ability to repeatedly rebuild lang/ghc from scratch? It was by far the most reliable failure that I found (that was a normal activity).] But I never observed the panics in normal operation, not even before the page-avoidance change was made. However, I was using 64-bit FreeBSD via Hyper-V with Windows 10 Pro as the boot OS. That might have caused automatic avoidance of the problematical page. (I also did a little VirtualBox activity instead of Hyper-V activity.) I tended to assign FreeBSD access to 14 hw-threads in Hyper-V, leaving 2 avoided for Windows 10 Pro availability. For the vast majority of things I'd left the BIOS at defaults. For example, no voltage changes of any kind --nor CPU frequency changes. AMD Cool' n' Quiet disabled, SVM Mode Enabled, Core C6 state disabled. 2nd memory profile used as-is. 2133 MT/s resulted. (It was not overclocking-style RAM.) And that is about it for adjustments. I no longer have access to that system. So that has limited what I managed to explore. It will probably be a month or so before I've access to a Ryzen system of some kind again. (A BIOS update failure also cut into the time I had on the Ryzen system: motherboard replacement.) --=20 You are receiving this mail because: You are the assignee for the bug.=
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-219399-8-wis3HAg1pb>