Arch Linux User Repository

There is a wiki page for AMDGPU PRO: https://wiki.archlinux.org/index.php/AMDGPU_PRO

Also it contains a link to bug report for amd hybrid graphics problem. Please vote if you are affected.

Yeah, I'm gonna call crap on the 20.10 EPB. Any attempt to use Blender to render using it, and my whole GPU locks up the PCIe bus. If I don't quickly switch to a TTY to kill Blender, my desktop will just lock up, period.

The 19.50 Orca driver works fine on my RX 480, though.

How many of the packages have you installed? I only installed:

amdgpu-core-meta
amdgpu-pro-core-meta
amdgpu-pro-libgl
lib32-amdgpu-pro-libgl
opencl-amdgpu-pro-orca
vulkan-amdgpu-pro
lib32-vulkan-amdgpu-pro

Using an RX 480. I’ve also kind of stopped using Linux for now, since within 20 minutes of booting I now get a pcie bus error with my gpu and kernel panic.

@kode54: Really? I'm also using gdm without any problems.

@naetherm I still can't make the libgl packages work with gdm, logging out of my desktop after installing them results in gdm crashing itself. The Vulkan drivers work, as do the Orca OpenCL drivers.

In the last days AMD released an early preview of 20.10 drivers, available here: https://www.amd.com/en/support/kb/release-notes/rn-rad-lin-20-10-early-preview

I've updated the PKGBUILD of this package. Installation is working fine for me. Available here: https://gist.github.com/naetherm/7a66839dc5b9a834c68b4d2e55fb8c23

But be careful when trying: Maybe I've missed something or have overlooked something.

Naetherm's PKGBUILD needed a bit of modification. Also, in my personal setup, I've removed the PKG_EXT override, since zstd compression has very little overhead on the process compared to the old default of xz.

I still can't make my desktop manager work or desktop log in with the libGL packages installed, so I've just decided to not bother with those for now.

In Ashark's PKGBUILD I noticed that the /19.50/ directory is missing in the source path. Could this be what is compromising the installation? Naetherm's PKGBUILD has it. Another difference is: backup=(etc/amd/amdapfxx.blb) Which in Naetherm is: backup=(etc/gbm/gbm.conf) Why the difference?

for what it's worth, @naetherm 's patched PKGBUILD worked fine for my purpose which was to use the AMFh.264 encoder in ffmpeg. 19.30 did not work on my 5600XT (encoded video was black).

19.50 won't work for me. Tried a few things but troubleshooting is witch-craft at best to me.

I had to modify the PKGBUILD you posted to get pacman to allow the install:

under

package_lib32-amdgpu-pro-libGL () {
<snip>
# extra_commands:

I added:

rm "${pkgdir}"/opt/amdgpu/share/drirc.d/10-amdgpu-pro.conf

Otherwise it came into conflict with the non-lib32 version because they both try to install that file.

Also have to delete some library symlink under /opt that's left behind after uninstalling the 19.30 packages. I forget which lib, pacman names it when I try installing with 'pacman -U *.pkg.tar'

After installation I did 'systemctl disable sddm' before rebooting so I'd get a terminal, which I'm sure isn't nmecessary but my luck with things like this tends to lean to catastrophe so I wanted to boot to a terminal for safety.

'systemctl start sddm' resulted in a crash. The only Xorg log I know to check(like I said, it's witchcraft to me), is /var/log/Xorg.0.log which the relevant bit seems to be(I'm running two GPUs BTW, the onboard intel and a Radeon RX 5700XT for a dual monitor setup):

[    29.547] (II) Initializing extension DRI2
[    29.549] (EE) intel(G0): sna_mode_shutdown_crtc: invalid state found on pipe 0, disabling CRTC:45
[    29.555] (EE) 
[    29.555] (EE) Backtrace:
[    29.555] (EE) 0: /usr/lib/Xorg (xorg_backtrace+0x4d) [0x556359ded76d]
[    29.555] (EE) 1: /usr/lib/Xorg (0x556359d3a000+0xa9cf8) [0x556359de3cf8]
[    29.555] (EE) 2: /usr/lib/libc.so.6 (0x7f331f5b7000+0x3bfb0) [0x7f331f5f2fb0]
[    29.555] (EE) 3: /usr/lib/libdrm_amdgpu.so.1 (amdgpu_bo_export+0x9e) [0x7f331e46204e]
[    29.555] (EE) 4: /usr/lib/x86_64-linux-gnu/dri/amdgpu_dri.so (0x7f32cca39000+0x1637b53) [0x7f32ce070b53]
[    29.555] (EE) 5: /usr/lib/x86_64-linux-gnu/dri/amdgpu_dri.so (0x7f32cca39000+0x13ab833) [0x7f32cdde4833]
[    29.555] (EE) 6: /usr/lib/x86_64-linux-gnu/dri/amdgpu_dri.so (0x7f32cca39000+0x13bd7dd) [0x7f32cddf67dd]
[    29.555] (EE) 7: /usr/lib/x86_64-linux-gnu/dri/amdgpu_dri.so (0x7f32cca39000+0x12317af) [0x7f32cdc6a7af]
[    29.555] (EE) 8: /usr/lib/x86_64-linux-gnu/dri/amdgpu_dri.so (eglCreateImageKHR+0x483) [0x7f32cdd33f33]
[    29.555] (EE) 9: /usr/lib/xorg/modules/libglamoregl.so (glamor_egl_create_textured_pixmap_from_gbm_bo+0x90) [0x7f331b3934b0]
[    29.555] (EE) 10: /usr/lib/xorg/modules/drivers/amdgpu_drv.so (0x7f331f780000+0x1bd7e) [0x7f331f79bd7e]
[    29.555] (EE) 11: /usr/lib/xorg/modules/drivers/amdgpu_drv.so (0x7f331f780000+0xee30) [0x7f331f78ee30]
[    29.555] (EE) 12: /usr/lib/Xorg (0x556359d3a000+0x67e9b) [0x556359da1e9b]
[    29.555] (EE) 13: /usr/lib/Xorg (0x556359d3a000+0x385b8) [0x556359d725b8]
[    29.555] (EE) 14: /usr/lib/libc.so.6 (__libc_start_main+0xf3) [0x7f331f5de153]
[    29.555] (EE) 15: /usr/lib/Xorg (_start+0x2e) [0x556359d733be]
[    29.555] (EE) 
[    29.555] (EE) Segmentation fault at address 0x10
[    29.555] (EE) 
Fatal server error:
[    29.555] (EE) Caught signal 11 (Segmentation fault). Server aborting
[    29.555] (EE) 
[    29.555] (EE) 
Please consult the The X.Org Foundation support 
     at <http://wiki.x.org>
 for help. 
[    29.555] (EE) Please also check the log file at "/var/log/Xorg.0.log" for additional information.
[    29.555] (EE) 
[    29.555] (II) AIGLX: Suspending AIGLX clients for VT switch
[    29.610] (EE) Server terminated with error (1). Closing log file.

The full log is at https://pastebin.com/ePvkQyfP

Sorry if that's not very helpful. I used to be better at tracking down issues like this but ADD doesn't let me stay on one hobby for very long and I forgot most of what I knew back then. :(

Let me know if there's any more info I could provide.

It's not working for me either, it doesn't start XOrg and I have to switch to the shell. Many errors "display not recognized" (and sddm), but without being able to determine the exact cause.