Fedora 37/36/35 NVIDIA Drivers Install Guide [530.30.02 / 525.89.02 / 520.56.06 / 515.86.01 / 510.108.03 / 470.161.03 / 390.157 / 340.108] - Comment Page: 66
This is guide, howto install NVIDIA proprietary drivers (manually using .run installer) on Fedora 37/36/35/34/33/32 and disable Nouveau driver. This guide works with GeForce 8/9/200/300/400/500/600/700/800/900/10/20/30/40 series cards.
GeForce RTX 40 series cards works with 530.xx, 525.xx, 520.xx NVIDIA drivers, (RTX 4090)
GeForce RTX 30 series cards works with 530.xx, 525.xx, 520.xx, 515.xx, 510.xx and 470.xx NVIDIA drivers, (RTX 3090, RTX 3080 and RTX 3070, RTX 3060, RTX 3060 Ti)
GeForce RTX 20 series cards works with 530.xx, 525.xx, 520.xx, 515.xx, 510.xx and 470.xx NVIDIA drivers (RTX 2080 Ti, RTX 2080, RTX 2070 Ti, RTX 2070, RTX 2060)
GeForce GT/GTX 600/700/800/900/10 series...
Hello!
Thank you for the very detailed instructions, unfortunately there must be a missing step for Fedora 28. I followed the instructions to the letter 3 separate times (after fresh installations) and was not able to get things to work correctly. I was able to get the Nvidia driver installed, but upon rebooting (into the graphical.target) the system hangs and GDM never comes up. I can confirm that the Nvidia kernel module was built correctly (using dkms status) and there’s nothing in the Xorg log files which indicates a problem.
My guess is there is something with the latest drivers (396.24) that doesn’t play nicely with Fedora 28. The only way I’ve gotten the Nvidia driver working with Fedora 28 is to use the Negativo17 packages (which are unfortunuately on 390.48).
Hello,
I have tested this on Fedora 28 several times and it’s working, so nothing is missing.
You didn’t tell anything about your hardware and you didn’t tell was nvidia module loaded correctly after boot? Because this could be also GDM problem?
Hi Robert,
Indeed, I had the same issues.
No specific messages during compilation except the GCC version mismatch.
I took the beta-driver : 396.18 BETA
This is just working very good !
I didn’t choose DKMS neither !!
Enjoy & success !
Yves
Hi, as I told you I have the same problem and I did clean install more than 4 times, I have the latest Fedora 28 with kernel 4.16.7-300, Asus GTX 1050 Ti, MD Asus x299 mark 1, CPU i9 7920x, I hope if anyone can help me.
Regards
Hi Meky,
Did you tried patched 390.48 installer or 396.24?
Or you can build nVidia correctly/without errors, but can’t boot to GDM?
Did you tried any other display manager?
I have 100% the same problem, please help.
390.48 also not working:
Error: Failed to run /usr/sbin/dkms build -m nvidia -v 390.48 -k 4.16.7-300 bla bla bla
Turns out it’s a known bug specifically with the PNY/Palit 1060 and the 396.xx drivers. Check out https://devtalk.nvidia.com/default/topic/1032725/linux/396-18-02-neon-sddm-crash-on-boot-xid-62-nvrm-rm_init_adapter-failed-for-device-bearing-min-/ for more information. Nvidia has apparently acknowleged the bug and has fixed it in 396.18.11 (beta). The fix should be in the next stable driver release. For now those of you using a PNY/Palit 1060 please stick with 390.48.
Thanks Robert for this information and link! Now I understand your problem better.
It worked!
nvidia-installer: version 396.24 ([email protected])
nvidia-installer: version 396.24 ([email protected])
01:00.0 VGA compatible controller: NVIDIA Corporation GK106 [GeForce GTX 660] (rev a1)
Lenovo P50 laptop with Fedora 28
I got stuck at “Starting switch root” after following the guide, and the display would flicker over and over on boot. Then finally let me drop to a terminal.
X was failing to start with 00 No screens found.
In order to resolve this, I had to go into my BIOS, Display settings, and switch Optimized graphics to “Discrete graphics only”
Apparently the Quatro M1000M on P50 has Optimus or Optimus-like technology, so disabling this solved the issue and I’m UP AND RUNNING!
Thanks for the guide and thanks all!
[root]# nvidia-installer -v | grep version
nvidia-installer: version 396.24 ([email protected]) Thu Apr 26 01:14:04 PDT 2018
[root]# uname -a
Linux serenity 4.16.6-202.fc27.x86_64 #1 SMP Wed May 2 00:09:32 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
[root]# lspci | grep -E “VGA|3D”
01:00.0 VGA compatible controller: NVIDIA Corporation GK106GLM [Quadro K2100M] (rev a1)
Your tutorial is working like a charm!
THANK YOU!!
(while the upgrade from Fedora 27 to Fedora 28 was a nightmare; destroyed the existing version, so I had to install 28 from scratch. what a mess!!!)
My specs:
# nvidia-installer -v |grep version
nvidia-installer: version 396.24 ([email protected]) Thu Apr 26
# uname -a
Linux 4.16.7-300.fc28.x86_64 #1 SMP Wed May 2 20:09:13 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
# lspci |grep -E “VGA|3D”
01:00.0 VGA compatible controller: NVIDIA Corporation GM206 [GeForce GTX 960] (rev a1)
On 4.17 kernel under Fedora 28, do I need to get patched nvidia drivers for it from somewhere or use stock drivers?
i was using 4.16 patched drivers before on kernel 4.16.X and worked fine but those fail on 4.17 ( so i boot back to 4.16 now)
Hi Viktor,
Haven’t done any testing with kernel 4.17 yet, but 4.16 should work.
Hi
Thank you for the excellent guides on how to install Nvidia Drivers on fedora.
I did a clean install of fedora 28, installed all the prerequisites and updatet every package to the newest version.
Then I tried to install the 396.24 driver from nvidia.
During installation I got an error for the dkms configuration. The problem was that my kernel 4.16.7-300 was compiled with gcc 8.0.1 and on my system I had gcc 8.1.1 installed. Therefore dkms failed.
For it to work I just donwgraded gcc and it’s dependencies with sudo dnf install gcc-8.0.1-0.20.fc28
Hi,
Normally next kernel build is released soon and this is not problem. Solution is downgrade GCC until next kernel is available. Or you can simply run following before installer:
I had a couple problems with Fedora 28, GeForce 8800 GTS, and the patched Nvidia 340 driver installation, but I was able to solve them. Maybe this will help someone:
1) After disabling nouveau and rebooting to non-graphical target, all I got was a black screen. The solution was to reboot again, go to the grub editor and add nomodeset option to the kernel boot line. After that I was able to get a character mode login and start installing the nvidia driver…
2) Within nvidia installer, I got an error saying that dkms build failed. After looking at the logs, the specific problem was that the version of gcc (8.0) that the current kernel (4.16.7-300.fc28.x86_64) was compiled with was different then the gcc version (8.1) available. There was no separate gcc 8.0 compatibility rpm package available in fedora repositories. The only solution I could think of was to do “dnf downgrade gcc-8.0.1”, which actually worked. I was able to restart the nvidia installer and complete the installation.
Hard to believe that Fedora 28 ships with gcc 8.1 and with a kernel that is compiled with gcc 8.0. Hope they start compiling kernels with gcc 8.1 soon.
Thanks!
Hi,
Normally next kernel build is released soon and this is not problem. Solution is downgrade GCC until next kernel is available. Or you can simply run following before installer:
I’ve been using your site for about a year. Only when new kernel families come out I need to re-do these steps. However, since 4.16.xx came out I have to re-do every time. I’m stuck. Thinking about installing gcc from testing (assuming they are using >7.3.1
Hi Randal,
What version of nVidia drivers and Fefora you are using?
I’ve installed the drivers successfully, can boot all the way up to gdm, login (graphical), but it all ends with a black screen and unresponsive system.
GP106 (GTX 1060 6GB)
Driver 396.24
Kernel 4.16.7-300.fc28.x86_64
‘dkms status’ shows success
Downgraded gcc to build on 8.0.1 during the install. also tried building on 8.1.1 with the ignore flag (mentioned elsewhere), same thing.
I’m at a loss how I can troubleshoot this?
Hi Jorn,
Could you post output of following command:
Yup! Works – works well and was easy to follow. Thanks!
nvidia-installer: version 390.48 ([email protected]) Thu Mar 22 01:07:44 PDT 2018
Linux 4.16.8-300.fc28.x86_64 #1 SMP Wed May 9 20:23:40 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
01:00.0 VGA compatible controller: NVIDIA Corporation GF119 [GeForce GT 520] (rev a1)
Thanks for the great guide! I’m new to Linux and it’s great to see support in the community.
I own Dell Xps 15 9650 with NVIDIA GTX 1050 (mobile) and Intel graphic cards. I use updated Fedora 28. And I use patched x86_64-390.48 driver.
I able to follow your guide exactly until
2.8.3. I don’t see nVidia Installer Accept License but after reading other comments I find out that it’s not a big deal.
2.8.7 I don’t see nVidia Installer Xorg Backup but I got “Installation of the kernel module for the NVIDIA Accelerated Graphic Driver for Linux-x86_64 (version 390.48) is now complete. No word about x configuration file.
Then I resumed to 2.9. After entered the command line, I rebooted the computer. When the computer boot up, it doesn’t display Fedora logo and anything else.
What I tried?
Reinstall the same version and different version. Same thing.
Edited the bootlines by appending 3 to end of linuxefi line. Same thing.
Boot in Fedora Rescue and I can see gui desktop but I still can’t see log in screen on regular Fedora.
I found out that my laptop has Optimus and I oversaw the first part. I tried BubmbleBee before and it didn’t work. I will retry it.
I don’t understand why Optimus work very well with Windows but not Linux. Why can’t NVIDIA take care of this?
Hello Alessandro,
Unfortunately there is no nVidias own support for Optimus devices on Linux. You can only run Cuda apps or do offscreen rendering. Bumblebee is one possible option, but looks like it’s dead project, because latest release is 5 years old.
Morning
Tutorial worked perfectly on Fedora 28 until the upgrade of the kernel to:
Linux localhost.localdomain 4.16.8-300.fc28.x86_64 #1 SMP Wed May 9 20:23:40 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
Now it crashes as soon as I plug in a monitor. I tried to re-install and the issue still exists. Any suggestions?
Thanks
Sorry forgot to say that I installed driver:
NVIDIA-Linux-x86_64-340.106-patched.run
Video Card:
01:00.0 VGA compatible controller: NVIDIA Corporation G92GLM [Quadro FX 2800M] (rev a2)
Thanks
Hello Percy,
Do you mean it crashes when you plug external monitor? Could you change another virtual console and check dmesg and journalctl outputs?
Hi JR
It seems to now crash intermittently regardless of whether I plug in a monitor or not. Often this is the case when the display turns off. I am reviewing logs and will update asap.
Thanks.
P
Errors in Journalctl:
— Logs begin at Thu 2018-05-17 22:09:37 SAST, end at Mon 2018-05-21 09:30:59 SAST. —
May 21 08:01:33 localhost.localdomain kernel: ACPI Error: Could not find/resolve named package element: TCNT (20180105/dspkginit-384)
May 21 08:01:33 localhost.localdomain kernel: ACPI Error: Could not find/resolve named package element: TCNT (20180105/dspkginit-384)
May 21 08:01:33 localhost.localdomain kernel: ACPI Error: Could not find/resolve named package element: TCNT (20180105/dspkginit-384)
May 21 08:01:33 localhost.localdomain kernel: ACPI Error: Could not find/resolve named package element: TCNT (20180105/dspkginit-384)
May 21 08:01:35 localhost.localdomain abrtd[1135]: ‘/var/spool/abrt/xorg-2018-04-27-15:33:30.619166-997’ is not a problem directory
May 21 08:01:36 localhost.localdomain abrtd[1135]: ‘/var/spool/abrt/xorg-2018-04-27-15:30:43.862796-921’ is not a problem directory
May 21 08:01:36 localhost.localdomain abrtd[1135]: ‘/var/spool/abrt/xorg-2018-05-02-21:10:41.519780-1145’ is not a problem directory
May 21 08:01:39 localhost.localdomain abrt-notification[2400]: System encountered a non-fatal error in __check_object_size()
May 21 08:01:44 localhost.localdomain spice-vdagent[3320]: Cannot access vdagent virtio channel /dev/virtio-ports/com.redhat.spice.0
May 21 08:03:02 localhost.localdomain spice-vdagent[14318]: Cannot access vdagent virtio channel /dev/virtio-ports/com.redhat.spice.0
May 21 08:03:06 localhost.localdomain kwin_x11[14189]: kwin_scene_opengl: GL_ARB_texture_non_power_of_two and GL_ARB_texture_rectangle missing
May 21 08:03:28 localhost.localdomain abrt-notification[17021]: Process 2307 (blueman-applet) of user 1000 encountered an uncaught GLib.GError exception
May 21 08:51:54 localhost.localdomain akonadiserver[14597]: org.kde.pim.akonadiserver: Control process died, committing suicide!
May 21 08:51:54 localhost.localdomain kglobalaccel5[28417]: Could not connect to any X display.
May 21 08:52:01 localhost.localdomain systemd-coredump[28426]: Process 5243 (java) of user 1000 dumped core.
Stack trace of thread 5292:
#0 0x00007f5a1adc0f4b raise (libc.so.6)
#1 0x00007f5a1adab591 abort (libc.so.6)
#2 0x00007f5a1a888104 _ZN9__gnu_cxx27__verbose_terminate_handlerEv (libjvm.so)
#3 0x00007f5a1a882f76 _ZN10__cxxabiv111__terminateEPFvvE (libjvm.so)
#4 0x00007f5a1a882fa3 _ZSt9terminatev (libjvm.so)
#5 0x00007f5a1a88335f __cxa_pure_virtual (libjvm.so)
#6 0x00007f5a1a6c547e _ZN12outputStream8print_crEPKcz (libjvm.so)
#7 0x00007f5a1a85b100 _ZN7VMError6reportEP12outputStream (libjvm.so)
#8 0x00007f5a1a85c7bc _ZN7VMError14report_and_dieEv (libjvm.so)
#9 0x00007f5a1a6bef4f JVM_handle_linux_signal (libjvm.so)
#10 0x00007f5a1a6b4ff3 _Z13signalHandleriP7siginfoPv (libjvm.so)
#11 0x00007f5a1b574fb0 __restore_rt (libpthread.so.0)
#12 0x00007f5a1adc3677 __run_exit_handlers (libc.so.6)
#13 0x00007f5a1adc37dc exit (libc.so.6)
#14 0x00007f5966ae62f3 gdk_x_io_error (libgdk-x11-2.0.so.0)
#15 0x00007f59fc3c7f62 _XIOError (libX11.so.6)
#16 0x00007f59fc3c5755 _XEventsQueued (libX11.so.6)
#17 0x00007f59fc3b71e0 XEventsQueued (libX11.so.6)
#18 0x00007f5a0541f3b6 n/a (n/a)
#19 0x00007f5a05007d00 n/a (n/a)
#20 0x00007f5a050082bd n/a (n/a)
#21 0x00007f5a05008302 n/a (n/a)
#22 0x00007f5a050007a7 n/a (n/a)
#23 0x00007f5a1a422ae6 _ZN9JavaCalls11call_helperEP9JavaValueP12methodHandleP17JavaCallArgumentsP6Thread (li>
#24 0x00007f5a1a422ff1 _ZN9JavaCalls12call_virtualEP9JavaValue11KlassHandleP6SymbolS4_P17JavaCallArgumentsP6>
#25 0x00007f5a1a423497 _ZN9JavaCalls12call_virtualEP9JavaValue6Handle11KlassHandleP6SymbolS5_P6Thread (libjv>
#26 0x00007f5a1a4becb0 _ZL12thread_entryP10JavaThreadP6Thread (libjvm.so)
#27 0x00007f5a1a80baa3 _ZN10JavaThread17thread_main_innerEv (libjvm.so)
#28 0x00007f5a1a80bbec _ZN10JavaThread3runEv (libjvm.so)
#29 0x00007f5a1a6baa28 _ZL10java_startP6Thread (libjvm.so)
#30 0x00007f5a1b56a564 start_thread (libpthread.so.0)
#31 0x00007f5a1ae8431f __clone (libc.so.6)
Stack trace of thread 5249:
#0 0x00007f5a1b5704fc [email protected]@GLIBC_2.3.2 (libpthread.so.0)
#1 0x00007f5a1a6b39c3 _ZN2os13PlatformEvent4parkEv (libjvm.so)
#2 0x00007f5a1a673d57 _ZN7Monitor5IWaitEP6Threadl (libjvm.so)
#3 0x00007f5a1a67468e _ZN7Monitor4waitEblb (libjvm.so)
#4 0x00007f5a1a3681b3 _ZN13GCTaskManager8get_taskEj (libjvm.so)
#5 0x00007f5a1a369048 _ZN12GCTaskThread3runEv (libjvm.so)
#6 0x00007f5a1a6baa28 _ZL10java_startP6Thread (libjvm.so)
#7 0x00007f5a1b56a564 start_thread (libpthread.so.0)
#8 0x00007f5a1ae8431f __clone (libc.so.6)
Stack trace of thread 5810:
#0 0x00007f5a1b5704fc [email protected]@GLIBC_2.3.2 (libpthread.so.0)
#1 0x00007f5a1a6b39c3 _ZN2os13PlatformEvent4parkEv (libjvm.so)
#2 0x00007f5a1a6a5865 _ZN13ObjectMonitor4waitElbP6Thread (libjvm.so)
#3 0x00007f5a1a4b6fb2 JVM_MonitorWait (libjvm.so)
#4 0x00007f5a05018107 n/a (n/a)
#5 0x00007f5a050082bd n/a (n/a)
#6 0x00007f5a050082bd n/a (n/a)
#7 0x00007f5a050082bd n/a (n/a)
#8 0x00007f5a050082bd n/a (n/a)
#9 0x00007f5a050082bd n/a (n/a)
#10 0x00007f5a05008302 n/a (n/a)
#11 0x00007f5a050007a7 n/a (n/a)
#12 0x00007f5a1a422ae6 _ZN9JavaCalls11call_helperEP9JavaValueP12methodHandleP17JavaCallArgumentsP6Thread (li>
#13 0x00007f5a1a422ff1 _ZN9JavaCalls12call_virtualEP9JavaValue11KlassHandleP6SymbolS4_P17JavaCallArgumentsP6>
#14 0x00007f5a1a423497 _ZN9JavaCalls12call_virtualEP9JavaValue6Handle11KlassHandleP6SymbolS5_P6Thread (libjv>
#15 0x00007f5a1a4becb0 _ZL12thread_entryP10JavaThreadP6Thread (libjvm.so)
#16 0x00007f5a1a80baa3 _ZN10JavaThread17thread_main_innerEv (libjvm.so)
#17 0x00007f5a1a80bbec _ZN10JavaThread3runEv (libjvm.so)
#18 0x00007f5a1a6baa28 _ZL10java_startP6Thread (libjvm.so)
#19 0x00007f5a1b56a564 start_thread (libpthread.so.0)
#20 0x00007f5a1ae8431f __clone (libc.so.6)
Stack trace of thread 5243:
#0 0x00007f5a1b56b9fd __GI___pthread_timedjoin_ex (libpthread.so.0)
#1 0x00007f5a1b358775 ContinueInNewThread0 (libjli.so)
#2 0x00007f5a1b353e9a ContinueInNewThread (libjli.so)
#3 0x00007f5a1b356f18 JLI_Launch (libjli.so)
#4 0x0000000000400696 main (java)
#5 0x00007f5a1adad1bb __libc_start_main (libc.so.6)
#6 0x0000000000400559 _start (java)
So Kernel was updated yesterday to 4.16.9-300.fc28.x86_64. I re-installed the driver after the update and the issue was resolved. Seems to have been an issue with the previous version.
In the uninstall guide, you might want to add to it an amendment to /etc/dnf/dnf.conf – specifically that it no longer contains the exclusion reference to the x11 drivers.
Hi Tom,
Actually no, because it shouldn’t be there when you follow install guide. I have two mention to remove it on install guide:
If you have exclude=xorg-x11* row in your /etc/dnf/dnf.conf, then you can remove it now and update xorg too.
If you have following row on /etc/dnf/dnf.conf file, then you can remove it:
Don’t like the look of next 10 lines. Have to login blind in multi-user.target and to startx and then it is OK barring couple of messages of xdg and booting.
[[email protected] ~]$ dmesg | tail
[ 66.034147] ? nvidia_frontend_unlocked_ioctl+0x3a/0x50 [nvidia]
[ 66.034149] ? do_vfs_ioctl+0xa4/0x610
[ 66.034150] ? __fput+0x12d/0x1f0
[ 66.034152] ? SyS_ioctl+0x74/0x80
[ 66.034154] ? do_syscall_64+0x74/0x180
[ 66.034157] ? entry_SYSCALL_64_after_hwframe+0x3d/0xa2
[ 66.034158] Code: 28 0c aa 41 51 4d 89 d8 48 c7 c0 80 05 0b aa 49 89 f1 48 89 f9 48 0f 45 c2 48 c7 c7 e0 28 0c aa 4c 89 d2 48 89 c6 e8 9d a1 e2 ff 0b 48 83 c4 18 c3 48 c7 c6 89 93 0d aa 49 89 f1 49 89 f3 eb
[ 66.034176] —[ end trace 0d702da9098fc0cb ]—
[ 66.530403] fuse init (API version 7.26)
[ 69.861839] rfkill: input handler disabled
[[email protected] ~]$ uname -r
4.16.8-300.fc28.x86_64
[[email protected] ~]$ nvidia-installer -v |grep version
nvidia-installer: version 390.59 ([email protected])
[[email protected] ~]$ uname -a
Linux localhost.localdomain 4.16.8-300.fc28.x86_64 #1 SMP Wed May 9 20:23:40 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
[[email protected] ~]$ lspci |grep -E “VGA|3D”
01:00.0 VGA compatible controller: NVIDIA Corporation GF119 [GeForce GT 610] (rev a1)
[[email protected] ~]$
Hi Aljoša,
If you check more your log, is it something like this:
https://devtalk.nvidia.com/default/topic/1031067/linux/-linux416-nvidia-390-48-nvidia_stack_cache-rip-0010-usercopy_warn-0x7e-0xa0/