AMDGPU drivers are a part of Linux itself, so you shouldn't need to install them manually. They're already there.
What Blender seems to want is ROCm HIP. The rocm-hip package might be what it wants? Try to install it and see if that works.
this post was submitted on 23 Jan 2024
28 points (96.7% liked)
Linux
48190 readers
904 users here now
From Wikipedia, the free encyclopedia
Linux is a family of open source Unix-like operating systems based on the Linux kernel, an operating system kernel first released on September 17, 1991 by Linus Torvalds. Linux is typically packaged in a Linux distribution (or distro for short).
Distributions include the Linux kernel and supporting system software and libraries, many of which are provided by the GNU Project. Many Linux distributions use the word "Linux" in their name, but the Free Software Foundation uses the name GNU/Linux to emphasize the importance of GNU software, causing some controversy.
Rules
- Posts must be relevant to operating systems running the Linux kernel. GNU/Linux or otherwise.
- No misinformation
- No NSFW content
- No hate speech, bigotry, etc
Related Communities
Community icon by Alpár-Etele Méder, licensed under CC BY 3.0
founded 5 years ago
MODERATORS
So, see the other comment for the dnf message when installing amdgpu-core. Everything else except amdgpu-core gets installed properly. The error says ERROR: This package can only be installed on EL8. Which makes no sense.
Did you add a repo for RHEL8 to your Fedora install? Please, undo that.
Please, don't blindly follow instructions you find online, particularly when it comes down to installing something as important as drivers.
Installing drivers from third party sources should be done only as the last resort and only if you know exactly what you're doing.
Ok, update. I removed some improper repos that the amdgpu-install had installed and I got things mostly working. Only problem is that rendering in Blender on GPU crashes it and I can’t seem to get good logs for the problem. I will try to get some logs and post them here.
I just realized this myself as well and posted it into an another comment. Should have replied here instead. But now the only problem is Blender crashing.
ghost_event_proc: ghost redraw decor 1
wm_event_do_handlers: Handling event
wmEvent type:1/LEFTMOUSE, val:1/PRESS, prev_type:1/LEFTMOUSE, prev_val:2/RELEASE, modifier={}, keymodifier:0, flag:{}, mouse:(1550,1038), utf8:'', pointer:0x7f696c2e7a80
UI_menutype_draw: opening menu "TOPBAR_MT_editor_menus"
UI_menutype_draw: opening menu "NODE_MT_editor_menus"
UI_menutype_draw: opening menu "FILEBROWSER_MT_editor_menus"
UI_menutype_draw: opening menu "VIEW3D_MT_editor_menus"
I0123 17:01:56.235076 4894 device.cpp:390] CPU render threads disabled for interactive render.
I0123 17:01:56.235121 4894 device.cpp:511] Mapped host memory limit set to 12,362,813,440 bytes. (11.51G)
I0123 17:01:56.235285 4894 device_impl.cpp:62] Using AVX2 CPU kernels.
Writing: /tmp/sky.crash.txt
Segmentation fault (core dumped)
[aapo@aapo-fedora ~]$ cat /tmp/sky.crash.txt
# Blender 3.6.7, Commit date: 1970-01-01 00:00, Hash unknown
bpy.ops.wm.open_mainfile(filepath="/home/aapo/Desktop/koodi/Blender/sky.blend", load_ui=True, use_scripts=False, display_file_selector=False, state=2) # Operator
bpy.ops.wm.tool_set_by_id(name="builtin.select_box", cycle=False, space_type='NODE_EDITOR') # Operator
bpy.ops.wm.tool_set_by_id(name="builtin.select_box", cycle=False, space_type='VIEW_3D') # Operator
bpy.ops.screen.userpref_show() # Operator
bpy.ops.wm.tool_set_by_id(name="builtin.select_box", cycle=False, space_type='NODE_EDITOR') # Operator
bpy.ops.wm.tool_set_by_id(name="builtin.select_box", cycle=False, space_type='VIEW_3D') # Operator
# backtrace
blender(+0x2dbff8c) [0x55b438d67f8c]
blender(+0x937963) [0x55b4368df963]
/usr/lib64/libc.so.6(+0x3dbb0) [0x7f699ea5fbb0]
# Python backtrace
[aapo@aapo-fedora ~]$
This is all the logs I seem to get. The logs before the cat command are from running Blender.
https://bugzilla.redhat.com/show_bug.cgi?id=2216594
https://gitlab.freedesktop.org/drm/amd/-/issues/2145
Looks like you're hitting a known bug that isn't fixed yet.
Reportedly, ROCm 5.6.1 is the lastest working version, so you could try to downgrade. Something dnf install rocm-*-5.6.1 should do the trick.
Well, I figured that out. But the problem is now my drivers are kind of broken. The error you are seeing in the log is what is not letting me download the drivers.
Ok, update. I removed some improper repos that the amdgpu-install had installed and I got things mostly working. Only problem is that rendering in Blender on GPU crashes it and I can't seem to get good logs for the problem. I will try to get some logs and post them here.
Also, if I basically remove everything, I will get this error:
`[aapo@aapo-fedora ~]$ amdgpu-install
[sudo] password for aapo:
ROCm 5.4 repository 154 kB/s | 208 kB 00:01
AMDGPU 5.4 repository 699 B/s | 548 B 00:00
Errors during downloading metadata for repository 'amdgpu':
- Status code: 404 for https://repo.radeon.com/amdgpu/5.4/rhel//main/x86_64/repodata/repomd.xml (IP: 13.82.220.49) Error: Failed to download metadata for repo 'amdgpu': Cannot download repomd.xml: Cannot download repodata/repomd.xml: All mirrors were tried Ignoring repositories: amdgpu Last metadata expiration check: 0:00:01 ago on Tue 23 Jan 2024 03:47:14 PM EET. Package amdgpu-lib-1:5.4.50400-1510348.el8.x86_64 is already installed. Package amdgpu-dkms-1:5.18.13.50400-1510348.el8.noarch is already installed. Error: Problem: package rocm-hip-runtime-5.4.0.50400-72.el8.x86_64 from rocm requires rocminfo = 1.0.0.50400-72.el8, but none of the providers can be installed
- conflicting requests
- nothing provides /usr/libexec/platform-python needed by rocminfo-1.0.0.50400-72.el8.x86_64 from rocm (try to add '--skip-broken' to skip uninstallable packages) [aapo@aapo-fedora ~]$ `
Here's also the dnf log:
[aapo@aapo-fedora ~]$ sudo dnf install amdgpu-core
AMDGPU 5.4 repository 588 B/s | 548 B 00:00
Errors during downloading metadata for repository 'amdgpu':
- Status code: 404 for https://repo.radeon.com/amdgpu/5.4/rhel//main/x86_64/repodata/repomd.xml (IP: 13.82.220.49)
Error: Failed to download metadata for repo 'amdgpu': Cannot download repomd.xml: Cannot download repodata/repomd.xml: All mirrors were tried
Ignoring repositories: amdgpu
Last metadata expiration check: 0:00:43 ago on Tue 23 Jan 2024 04:14:52 PM EET.
Dependencies resolved.
================================================================================
Package Arch Version
Repository Size
================================================================================
Installing:
amdgpu-core noarch 1:5.4.50400-1510348.el8
repo.radeon.com_amdgpu_5.4_rhel_8.7_main_x86_64_ 8.1 k
Transaction Summary
================================================================================
Install 1 Package
Total download size: 8.1 k
Installed size: 0
Is this ok [y/N]: y
Downloading Packages:
amdgpu-core-5.4.50400-1510348.el8.noarch.rpm 19 kB/s | 8.1 kB 00:00
--------------------------------------------------------------------------------
Total 19 kB/s | 8.1 kB 00:00
Running transaction check
Transaction check succeeded.
Running transaction test
Transaction test succeeded.
Running transaction
Preparing : 1/1
Running scriptlet: amdgpu-core-1:5.4.50400-1510348.el8.noarch 1/1
ERROR: This package can only be installed on EL8.
error: %prein(amdgpu-core-1:5.4.50400-1510348.el8.noarch) scriptlet failed, exit status 1
Error in PREIN scriptlet in rpm package amdgpu-core
Verifying : amdgpu-core-1:5.4.50400-1510348.el8.noarch 1/1
Failed:
amdgpu-core-1:5.4.50400-1510348.el8.noarch
Error: Transaction failed
[aapo@aapo-fedora ~]$