Unusual behaviour with dual monitors with RX 5600 XT
System information
System: Host: maxwell-pc Kernel: 5.7.0-1-MANJARO x86_64 bits: 64 compiler: gcc v: 9.3.0 Desktop: KDE Plasma 5.18.5
tk: Qt 5.14.2 wm: kwin_x11 dm: SDDM Distro: Manjaro Linux
CPU: Topology: 8-Core model: AMD Ryzen 7 2700X bits: 64 type: MT MCP arch: Zen+ rev: 2 L2 cache: 4096 KiB
flags: avx avx2 lm nx pae sse sse2 sse3 sse4_1 sse4_2 sse4a ssse3 svm bogomips: 118237
Speed: 1884 MHz min/max: 2200/3700 MHz Core speeds (MHz): 1: 1906 2: 2212 3: 2079 4: 1971 5: 1910 6: 1916 7: 4250
8: 4300 9: 2186 10: 1869 11: 2016 12: 1960 13: 1933 14: 1808 15: 1910 16: 1915
Graphics: Device-1: Advanced Micro Devices [AMD/ATI] Navi 10 [Radeon RX 5600 OEM/5600 XT / 5700/5700 XT] vendor: Gigabyte
driver: amdgpu v: kernel bus ID: 08:00.0 chip ID: 1002:731f
Display: x11 server: X.Org 1.20.8 driver: amdgpu FAILED: ati unloaded: modesetting,radeon alternate: fbdev,vesa
compositor: kwin_x11 resolution: 1920x1080~144Hz, 1920x1080~144Hz
OpenGL: renderer: AMD Radeon RX 5600 XT (NAVI10 DRM 3.37.0 5.7.0-1-MANJARO LLVM 10.0.0) v: 4.6 Mesa 20.0.6
direct render: Yes
The issue
For some reason, after my two monitors turn off due to inactivity, per my power management settings (which I disabled for the time being due to the following problem), and I reactivate them to do something on the computer, my GPU's performance becomes severely gimped. E.g. normally I can play Deep Rock Galactic at 144 FPS easily, but after the GPU's performance is affected, it runs at 10 FPS at the very best. This affects any application that requires graphical processing.
For the longest time I was unable to tell what the problem was, until by chance I discovered that turning off and on my secondary monitor resolves the problem, instantly. Turning off the secondary monitor does not resolve the issue, but turning it back on again does. I don't know why that is, but something about two monitors resuming operation simultaneously sometimes negatively affects the GPU until one of them is turned off and on again. I am unsure if this is a problem with Mesa drivers or something with the GPU in itself.
Sometimes when the computer is idle for long enough, even with power management settings disabled completely, this issue reoccurs. Other things cause this issue to manifest itself immediately, with slight variation, such as running inxi --full --verbosity=7 --filter --no-host
(that I ran to show in bug tracking for my Linux distro). If I run the aforementioned command, the desktop environment has tremendous problems with switching windows, as it freezes for at least five seconds any time it attempts to switch windows. I know this is related to this same issue, because, again, restarting the secondary monitor alleviates the issue.
Regression
This problem used not to occur when I used the same computer, but with an NVIDIA GTX 1060 instead of an RX 5600 XT GPU.
Possibly relevant error message
[39810.377858] xhci_hcd 000:09:00.3: Host halt failed, -110
[39812.684277] amdgpu: [powerplay] failed send message: DisableAllSmuFeatures (7) oparam: 0x00000000 response 0xffffffc2
[39812.684300] [drm:amdgpu_device_ip_suspend_phase2 [amdgpu]] *ERROR* suspend of IP block <smu> failed -62
[xxxxx.xxxxxx] amdgpu: [powerplay] Msg issuing pre-check failed and SMU may be not in the right state!
[xxxxx.xxxxxx] amdgpu: [powerplay] [PrepareMp1] Failed!
[xxxxx.xxxxxx] [drm:amdgpu_device_ip_suspend_phase [amdgpu]] *ERROR* SMU failed to set mp1 state 2, -68
[xxxxx.xxxxxx] reboot: Power down
(The above message might contain some typos due to poor image capture with my phone)
Attachments
dmesg_output.txt dmidecode.log glxinfo.log lshw.log lsmod.amdgpu.log lspci.nn.log lspci.vnn.log modinfo.amdgpu.log Xorg.0.log