I can't use my second AMD GPU in my laptop

Ah that sux. I won’t be able to play upcoming Star Wars: Lego Skywalker Saga with HD 520

I’m thinking you’re having a hardware problem by now…

  • Is the GPU still under warranty?
  • Do you know someone you can swap GPUs with? (same model)

No it’s not under warranty.
And my gpu is combined with motherboard.

What kind of games? Proton or native?

Also please post the kernel messages:

sudo dmesg

Any kind of games, apps also i tested with reflect command. Tried Proton Experimental, 6.36, Steam linux compatibility and native.
The output was too long. So i give you the link of the output

I have searched a bit and and it seems it can be a lucky end…

Your card:

Is a GCN 1 card and therefore you should be able to run it with the amdgpu driver instead of radeon.

Try this solution:

That force to load amdgpu first and set the necessary options for older gpus which have support for radeon and amdgpu. But it is still called experimental, although its pretty stable and faster on most machines with older still compatible gpus.

This package does the same thing:

pamac install amdgpu-experimental
1 Like

omg im now using amdgpu driver (which i used before) but i didn’t apply that mkinitcpio -P now it shows R5 M330 in xrandr --listproviders instead of Hainan thing. Unlucky steam gave an error You are missing the following 32-bit libraries, and Steam may not run: libc.so.6 I will search this if i fix i will inform you about game status. (Note: i am able to run Pycharm with DRI_PRIME=1)

You have to enable the multilib repo and install lib32-glibc … i guess you will need to install more lib32 packages.

in /usr/lib32/ i have libc.so.6 but dunno why system ignores it

Well i fixed steam, but DRI_PRIME=1 still crashes

@DavutHaxor

  1. Then please run steam in a terminal and post the output here when it crashes.
  2. Also for Proton Games, add there atthe startup options: PROTON_LOG=1 PROTON_LOG_DIR=$HOME %command% and post the log here.
  3. Also sudo dmesg.
  4. journalctl --boot 0 --no-pager

Normally there are hints why it crashes.

it freezes my computer so i dont know how can i post the terminal output of steam, i will try

Execute:

steam > ~/Documents/Steam.log

after a REISUB, that file will still be there…

:crossed_fingers:

ok then…

Add at then startup options: PROTON_LOG=1 PROTON_LOG_DIR=$HOME %command% and post the log here. This will write the log at your home dir.

If freezes completely now. After a restart type this:

journalctl --boot -1 --no-pager
journalctl --boot -1 --dmesg --no-pager

0 → current boot
-1 → previous boot

Run steam like @Fabby suggested or do it like that:

steam | tee -a ~/steam_log.txt

The log should be available even if it freezes or you REISUB or hard reset.

Then post everything.

FYI: hard reset on an EXT4 volume risks the file NOT being there as it’s a journaling file system and might roll back the last journal log…

:man_shrugging:

Proton gives me error so i run normal and Steam Linux Support tool.
CSGO didn’t freeze my system but i got 20 fps with

DRI_PRIME=1 %command%

Steam.log (hastebin)

journalctl with 0 (hastebin)

EDIT: I think i can’t use my second gpu anymore. Today when i opened my pc it showed 103 updates from pacman. I installed them and restarted my pc. DRI_PRIME=1 is useless now.
EDIT2: Now i can use DRI_PRIME, i will send up to date outputs
EDIT3: I updated outputs. I saw some amdgpu errors. You can examine the last lines.

@DavutHaxor

Looks like it have only 256M VRAM:

Eyl 16 18:49:02 davut-laptop kernel: amdgpu 0000:01:00.0: PCIE GART of 256M enabled (table at 0x000000F400000000).

But Csgo need at least over 256M. So I guess the timeout here:

Eyl 16 18:49:26 davut-laptop kernel: [drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring gfx timeout, signaled seq=527, emitted seq=528

is because of less VRAM.

it has 2048 mb vram and also it can’t open apps like firefox, scrcpy, discord, steam too

I can’t really help here. I see a lot of issues regarding this error on Mesa Issues.

I have not AMD GPU for testing, but there are some tweaks you can do: drm/amdgpu AMDgpu driver — The Linux Kernel documentation

Example:

  1. Lets say our want to turn off dynamic power management
  2. Then add amdgpu.dpm=0 to the kernel parameter
    3 OR add it to /etc/modprobe.d/ with a new file:
options amdgpu dpm=0

I guess you have to tinker there since it is experimental for older GPUs.

Use Kernel parameters
amdgpu.gartsize=0
amdgpu.runpm=0
amdgpu.vramlimit=0
amdgpu.dc=1
amdgpu.dpm=0
And tell me if it works