This may be EFI bios related / amdgpu.ko kernel driver dpm powerplay related. The dmesg.txt attached seems to confirm this in this ticket. The problem is not software, it's overheating / hardware related. It's affecting AMI EFI bios code + AMD GPU SBIOS powerplay voltage tables. See this thread. https://forum-en.msi.com/index.php?topic=298468.0 Here is an email I just sent to msi support about this saga, as they are trying to replicate this defect as it's no good when fans either don't spin up on some cards, or stay locked at 1000rpm out of 3000rpm. No good = the card crashing corruption etc software issues. This user needs to monitor his fan speed / temperature when using the GPU. I have emails about how to that as well if he wants. I'm using sabayon.org Linux out of the box to reproduce the error. Latest version I think 18.02 or 17.xx from memory. ╠ @@ Package: sys-kernel/linux-sabayon-4.14.12-r1 branch: 5, [sabayon-weekly] ╠ Available: version: 4.14.12-r1 ~ tag: NoTag ~ revision: 0 ╠ Installed: version: 4.14.12-r1 ~ tag: NoTag ~ revision: 0 ╠ Slot: 4.14 ╠ Homepage: https://github.com/Sabayon/kernel ╠ Description: Official Sabayon Linux Standard ╠ kernel image But i'm moving to pure gentoo.org custom compiled everything. Right now whilst i'm transitioning I"m using an out of the box kernel. All it matters is if upstream Linux is broken then many linux distributions by default are broken. And this admgpu driver not handling modified voltage tables is going to kill cards running linux. Ubuntu's kernel may have a custom patch to fix this already, that's why you can't replicate it now. let me look at the sources / patches for ubuntu kernel now. https://bugs.freedesktop.org/show_bug.cgi?id=100443 See the original post I put up, that's other people getting it. Someone else on the internet with the bug in Ubuntu: http://www.cadalyst.com/%5Blevel-1-with-primary-path%5D/rx-850-ubuntu-driver-problems-34295 " RX 580 Ubuntu Driver Problems Wed, 05/31/2017 - 00:40 — Anonymous I just got a pair of XFX RX 580 card and am having some trouble with the amdgpu-pro drivers. On a fresh install of Ubuntu 16.04.2, after updating packages, updating the kernel to 4.8, installing the AMDGPU-PRO 17.10 drivers, and rebooting, I see these messages in the log " Maybe it's been patched in more recent ubuntu versions. The goal is to work around the bug in firmware so people who run older linuxes / unpatched don't get hit. Anything to mitigate this overheating issue either at the linux kernel or firmware is a good move. He used 16.04.2 without a kernel update from the CD. But Ubuntu 16.04.2 has other problems.... znmeb commented on 2017-12-04 23:05 This software doesn't even work on Ubuntu 16.04.3 LTS, which AMD supposedly supports!! See http://support.amd.com/en-us/kb-articles/Pages/AMDGPU-PRO-Driver-Compatibility-Advisory-with-Ubuntu-16.04.2-and-16.04.3.aspx I couldn't even make it work on 16.04.2. I've pretty much given up on AMD. But could be that.... Maybe they are putting the spin on the powerplay issue as something else. Remember AMDGPU-Pro isn't reconmended for most of the AMD GPU chipsets you sell. AmdGPU "all open" they also bundle with ubuntu is just their own unmodified but compiled version pure opensource amdgpu.ko kernel driver, linux-firmware for the polaris10-11, mesa opengl - vaapi etc You want to do your testing on VANILLA Linux kernels. I .e. learn how to compile a vanilla kernel in ubuntu or use vanilla kernel source. That's what all the other non-ubuntu linux distributions are using. They have ready pre-cooked ones for you here: https://wiki.ubuntu.com/Kernel/MainlineBuilds Does it break with any of the < 4.14 Linux kernels from there? Maybe Ubuntu's patched it already. I'll have a quick look on their bug tracker to see if I can find it..... <10 mins later.... > Hey... I just found someone else yet again who video card was crashing.... Who was running Ubuntu... who had no fan control.. Using the ubuntu bug database. I told you it's a global problem. All these errors with the video cards crashing at high res, in games in videos etc is OVERHEATING issues being reported to linux bugtrackers as "software issues". It's a mess... https://bugs.launchpad.net/ubuntu/+source/xorg/+bug/1740484 Look at the dmesg.txt attached to that bug... the error message I kept raving about about powerplay votlage tables... https://launchpadlibrarian.net/351532328/CurrentDmesg.txt He's also got EFI firmware from AMI like you guys use, but not an MSI card. It's AMI EFI firmware + messed voltage tables + linux kernel = dead powerplay no fan control I think. That's me searching the UBUNTU bugs database for 10 mins guys so you know it's real. Try using an older motherboard like the same as mine with OLDER AMI EFI bios driver version. That EFI version will be right at the top of the dmesg. Details on my motherboard etc are in the original post. Yo [ 0.000000] efi: EFI v2.60 by American Megatrends [ 0.000000] efi: ACPI 2.0=0xd48df000 ACPI=0xd48df000 SMBIOS=0xdb92f000 SMBIOS 3.0=0xdb92e000 ESRT=0xd815b998 [ 0.000000] random: fast init done [ 0.000000] SMBIOS 3.0.0 present. [ 0.000000] DMI: System manufacturer System Product Name/PRIME B350M-A, BIOS 3401 12/04/2017 Finding exactly what is required to replicate this at your end will be useful. Just search amdgpu in the ubuntu bug ticket to see other users who are suffering. https://bugs.launchpad.net/ubuntu?field.searchtext=amdgpu&search=Search&field.status%3Alist=NEW&field.status%3Alist=INCOMPLETE_WITH_RESPONSE&field.status%3Alist=INCOMPLETE_WITHOUT_RESPONSE&field.status%3Alist=CONFIRMED&field.status%3Alist=TRIAGED&field.status%3Alist=INPROGRESS&field.status%3Alist=FIXCOMMITTED&field.assignee=&field.bug_reporter=&field.omit_dupes=on&field.has_patch=&field.has_no_package = Cheers, Luke Среда, 24 января 2018, 16:10 +07:00 от MSI OCSS