You are not logged in.

#1 2019-04-21 00:09:26

pontiaj2066
Member
Registered: 2019-04-21
Posts: 16

System completely locks up after gaming for a while with Vega 56 GPU

The only thing I can do is force shut it off when it freezes. (Just learned about Sysreq REISUB, so I'll do that from now on) It's happened in Alien: Isolation and Slime rancher, but not Factorio, Terraria, or Rimworld. I'm using 4.19.12-arch1-1-vfio, but it happens with the normal kernel as well. Intel i5-8600K, AMD Vega 56, 16GB Ram, 8 Swap. I monitored the temperatures while I played. GPU was at 56, CPU 75. I figured it was a GPU issue, so I did journalctl -b-1 | grep amdgpu and I got this: https://pastebin.com/csviFFgf. Normal journalctl just adds wifi related stuff. Anyone know what to do?

Last edited by pontiaj2066 (2019-04-23 23:54:18)

Offline

#2 2019-04-21 06:25:43

seth
Member
Registered: 2012-09-03
Posts: 49,951

Re: System completely locks up after gaming for a while with Vega 56 GPU

Offline

#3 2019-04-21 15:29:56

pontiaj2066
Member
Registered: 2019-04-21
Posts: 16

Re: System completely locks up after gaming for a while with Vega 56 GPU

Yeah. I just tried the 5.0.8 kernel and it happened again

Offline

#4 2019-04-23 02:34:18

pontiaj2066
Member
Registered: 2019-04-21
Posts: 16

Re: System completely locks up after gaming for a while with Vega 56 GPU

Also happens on the 4.18 kernel, which someone told me should help, but it didn't

Offline

#5 2019-04-23 09:10:40

loqs
Member
Registered: 2014-03-06
Posts: 17,192

Re: System completely locks up after gaming for a while with Vega 56 GPU

You could try 5.1-rc6 or linux-amd-staging-drm-next-git.

Offline

#6 2019-04-23 10:57:48

r0b0t
Member
From: /tmp
Registered: 2009-05-24
Posts: 505

Re: System completely locks up after gaming for a while with Vega 56 GPU

pontiaj2066 wrote:

The only thing I can do is force shut it off when it freezes. (Just learned about Sysreq REISUB, so I'll do that from now on) It's happened in Alien: Isolation and Slime rancher, but not Factorio, Terraria, or Rimworld. I'm using 4.19.12-arch1-1-vfio, but it happens with the normal kernel as well. Intel i5-8600K, AMD Vega 56, 16GB Ram, 8 Swap. I monitored the temperatures while I played. GPU was at 56, CPU 75. I figured it was a GPU issue, so I did journalctl -b-1 | grep amdgpu and [I got this](https://pastebin.com/csviFFgf). Normal journalctl just adds wifi related stuff. Anyone know what to do?

Only after gaming?
I read this and just thought let me try switch to my radeon rx 590, as soon as I plugin the HDMI

 kernel: [drm] SADs count is: -2, don't need to read it

and it's extremely laggish , even the mouse isn't moving properly...
BTW: no games installed, no heavy graphic applied...

amdgpu seems to need some serious work done...

Offline

#7 2019-04-23 16:52:01

Lone_Wolf
Member
From: Netherlands, Europe
Registered: 2005-10-04
Posts: 11,868

Re: System completely locks up after gaming for a while with Vega 56 GPU

The intel i5-8600K comes with an igp, making  pontiaj2066 system a hybrid graphics setup.

Pontaj2066, did you use the PRIME wiki page to setup your system ?


Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.


(A works at time B)  && (time C > time B ) ≠  (A works at time C)

Offline

#8 2019-04-23 23:53:15

pontiaj2066
Member
Registered: 2019-04-21
Posts: 16

Re: System completely locks up after gaming for a while with Vega 56 GPU

r0b0t wrote:

Only after gaming?

Only while in the middle of gaming. Specifically 3d games.

Lone_Wolf wrote:

Pontaj2066, did you use the PRIME wiki page to setup your system ?

No, I didn't. It appears to be for laptops rather than desktops though

loqs wrote:

You could try 5.1-rc6 or linux-amd-staging-drm-next-git.

Is 5.1-rc6 the latest kernel? I'm only seeing 5.0.8 when I try to update. I'll try the amd staging kernel

Offline

#9 2019-04-24 07:31:03

seth
Member
Registered: 2012-09-03
Posts: 49,951

Re: System completely locks up after gaming for a while with Vega 56 GPU

It appears to be for laptops rather than desktops though

lspci
glxinfo

(Though the log suggests the AMD GPU to be in use and you probably have an output wired to it and an according Xorg config?)

Offline

#10 2019-04-24 10:07:46

Lone_Wolf
Member
From: Netherlands, Europe
Registered: 2005-10-04
Posts: 11,868

Re: System completely locks up after gaming for a while with Vega 56 GPU

pontiaj2066 wrote:
Lone_Wolf wrote:

Pontaj2066, did you use the PRIME wiki page to setup your system ?

No, I didn't. It appears to be for laptops rather than desktops though

All recent x86 systems with an integrated gpu use Hybrid Graphics muxless design.
It's a processor attribute that doesn't depend on formfactor, size , processor socket etc .

When you use a hybrid graphics processor and a discrete videocard with opensource X drivers, you are running a PRIME setup.


Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.


(A works at time B)  && (time C > time B ) ≠  (A works at time C)

Offline

#11 2019-04-25 00:04:31

pontiaj2066
Member
Registered: 2019-04-21
Posts: 16

Re: System completely locks up after gaming for a while with Vega 56 GPU

seth wrote:
lspci
glxinfo

(Though the log suggests the AMD GPU to be in use and you probably have an output wired to it and an according Xorg config?)

Here's the output of lspci and glxinfo. I've got two outputs wired to the graphics card. I never made an Xorg config, and /etc/X11/xorg.conf.d/ is empty, so I don't believe I have one

Lone_Wolf wrote:

When you use a hybrid graphics processor and a discrete videocard with opensource X drivers, you are running a PRIME setup.

Oh, I didn't know that. Looking through the PRIME wiki page, I made sure I have the xf86-video-amdgpu package, and

xrandr --listproviders 

returns:

Providers: number : 1
Provider 0: id: 0x59 cap: 0xf, Source Output, Sink Output, Source Offload, Sink Offload crtcs: 6 outputs: 4 associated providers: 0 name:Radeon RX Vega @ pci:0000:03:00.0

and

glxinfo | grep "OpenGL renderer"

returns:

OpenGL renderer string: Radeon RX Vega (VEGA10, DRM 3.27.0, 4.19.12-arch1-1-vfio, LLVM 8.0.0)

so it looks like only the dedicated GPU is in use

Offline

#12 2019-04-25 07:17:09

seth
Member
Registered: 2012-09-03
Posts: 49,951

Re: System completely locks up after gaming for a while with Vega 56 GPU

interestingly

04:00.0 VGA compatible controller: NVIDIA Corporation GP104 [GeForce GTX 1070] (rev a1)

but no intel vga device (unless I need glasses - I assume it will show up if you remove both dedicated GPUs?)

What's the status of the nvidia GPU, is it unpowered? Does it show up in your xorg log?
For a wild guess you could try to remove it, maybe even swap PCI slots -  consult the mainboards manual for pot. preferred PCIe lanes.

Can you knock out the GPU w/ synthetic tests like https://aur.archlinux.org/packages/gputest/ and/or https://aur.archlinux.org/packages/glmark2/

Edit: apparently better use https://aur.archlinux.org/packages/glmark2-git/

Last edited by seth (2019-04-25 07:18:42)

Offline

#13 2019-04-25 10:51:04

Lone_Wolf
Member
From: Netherlands, Europe
Registered: 2005-10-04
Posts: 11,868

Re: System completely locks up after gaming for a while with Vega 56 GPU

OpenGL renderer string: Radeon RX Vega (VEGA10, DRM 3.27.0, 4.19.12-arch1-1-vfio, LLVM 8.0.0)

It also shows this system is not using latest kernel which can be cause of problems for amd vega cards.
I can't find a 4.19.12 vfio kernel in AUR, which aur package are you using ?

The vfio kernel is mostly used in special cases to circumvent uefi firmware iommu issues.
pontiaj2066, post full dmesg/journal please.

What is the brand and model of your motherboard ?
apart from an amd and nvidia video card you also seem to have a video capture card, am I correct ?

What is the kernel commandline executed by your bootloader ?
are there any files in /etc/modprobe.d/ and if so, what's their content ?

Is your uefi firmware the latest supplied by the manufacturer ?


Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.


(A works at time B)  && (time C > time B ) ≠  (A works at time C)

Offline

#14 2019-04-25 11:35:32

loqs
Member
Registered: 2014-03-06
Posts: 17,192

Re: System completely locks up after gaming for a while with Vega 56 GPU

4.19.12-arch1-1-vfio could be https://aur.archlinux.org/cgit/aur.git/ … 56b13d895e although compiling with LLVM instead of GCC would be custom.

Offline

#15 2019-04-25 12:21:19

seth
Member
Registered: 2012-09-03
Posts: 49,951

Re: System completely locks up after gaming for a while with Vega 56 GPU

Comments #3 & #4 suggested it's not the particular kernel, though *shrug*

Offline

#16 2019-04-26 00:57:04

pontiaj2066
Member
Registered: 2019-04-21
Posts: 16

Re: System completely locks up after gaming for a while with Vega 56 GPU

I tried again on the 5.0.9 kernel. Here's the full journalctl part 1 and part 2, along with the full dmesg. I had used the 4.19 vfio kernel for GPU passthrough because my motherboard IOMMU groupings had both GPUs in the same group. 4.19 was the current one found here at the time, and I'm not able to update it do to this error. I had the GTX 1070 separated using the GPU passthrough page found here. I do have a capture card. It's the Magewell Pro Capture HDMI card. My motherboard is the ASUS ROG Strix Z370-E. Its firmware is up to date. /etc/modprobe.d contains 3 files: firewalld-sysctls.conf, vfio.conf, and ProCapture.conf. Their contents are here. I believe /boot/grub//grub.cfg is the command line executed, but I'm not sure. Now I'll see if the GPU test will freeze it too, and I'll try without the Nvidia card in

Offline

#17 2019-04-27 19:01:43

pontiaj2066
Member
Registered: 2019-04-21
Posts: 16

Re: System completely locks up after gaming for a while with Vega 56 GPU

I ran the GPU test overnight, which didn't freeze it, and playing without the GTX 1070 still froze it

Offline

#18 2019-04-27 19:04:47

seth
Member
Registered: 2012-09-03
Posts: 49,951

Re: System completely locks up after gaming for a while with Vega 56 GPU

Can you run the game on the nvidia chip?
Can you run the game on an openbox session? (plasma + kwin open a ton of GL contexts)

Offline

#19 2019-04-27 19:14:43

loqs
Member
Registered: 2014-03-06
Posts: 17,192

Re: System completely locks up after gaming for a while with Vega 56 GPU

linux-vfio does not build due to https://bugs.archlinux.org/task/62332
download https://git.archlinux.org/svntogit/pack … fbd25223fd and apply it with git apply -v -p2 cfe52e9aa8168d9571bedf8a376e2cfbd25223fd.patch

Offline

#20 2019-04-27 19:45:57

pontiaj2066
Member
Registered: 2019-04-21
Posts: 16

Re: System completely locks up after gaming for a while with Vega 56 GPU

seth wrote:

Can you run the game on the nvidia chip?

Yes, it appeared to work(tried for around an hour and a half before I stopped playing, which is longer than the AMD card has lasted), but I'd prefer to use the AMD card

Can you run the game on an openbox session? (plasma + kwin open a ton of GL contexts)

I'm not too sure what this means. I am running plasma + kwin. If you mean having a ton of windows open, then yes, it freezes. It freezes no matter how many I have open while playing.

loqs wrote:

linux-vfio does not build due to https://bugs.archlinux.org/task/62332
download https://git.archlinux.org/svntogit/pack … fbd25223fd and apply it with git apply -v -p2 cfe52e9aa8168d9571bedf8a376e2cfbd25223fd.patch

Thanks, I'll try that!

Offline

#21 2019-04-27 19:55:02

seth
Member
Registered: 2012-09-03
Posts: 49,951

Re: System completely locks up after gaming for a while with Vega 56 GPU

I mean to install https://www.archlinux.org/packages/comm … 4/openbox/ and run that instead of plasma

Offline

#22 2019-04-27 22:16:18

pontiaj2066
Member
Registered: 2019-04-21
Posts: 16

Re: System completely locks up after gaming for a while with Vega 56 GPU

I just tried that, and it froze. But this time, I was greeted with this after it was frozen for a few seconds

Offline

#23 2019-04-27 22:42:22

pontiaj2066
Member
Registered: 2019-04-21
Posts: 16

Re: System completely locks up after gaming for a while with Vega 56 GPU

The accompanying journalctl from that freeze is here

Offline

#24 2019-04-28 07:00:19

seth
Member
Registered: 2012-09-03
Posts: 49,951

Re: System completely locks up after gaming for a while with Vega 56 GPU

At least you don't seem to be alone…
https://www.phoronix.com/scan.php?page= … -19.04-Bug

Offline

#25 2019-04-28 17:16:54

pontiaj2066
Member
Registered: 2019-04-21
Posts: 16

Re: System completely locks up after gaming for a while with Vega 56 GPU

I'm starting to get the impression that the only fix is waiting for a driver patch, or using Nvidia until then

Offline

Board footer

Powered by FluxBB