Fedora 28/27/26 nVidia Drivers Install Guide - Comment Page: 64
This is guide, howto install nVidia proprietary drivers on Fedora 28/27/26/25/24/23/22/21 and disable Nouveau driver. This guide works with GeForce 6/7/8/9/200/300/400/500/600/700/800/900/10 series cards.
GeForce 400/500/600/700/800/900/10 series cards works with 390.xx and 396.xx nVidia drivers
GeForce 8/9/200/300 series cards works with 340.xx nVidia drivers
GeForce 6/7 series cards works with 304.xx nVidia drivers
This is totally different guide than my earlier guides, like Fedora 20 nVidia driver install. This guide uses nVidia drivers directly from nVidia site and dkms to help on kernel updates. Have to say that I have had and heard strange problems with different repos...
I am having a problem with step 2.7. Whenever i try and run ./NVIDIA-Linux-*.run i get “permission denied”.
Hi Darkon,
Did you chmod +x that file first (step. 2.2 Make nVidia installer executable)?
https://imgur.com/lYEQwRK
nvidia-installer: version 390.48 ([email protected])
Linux localhost.localdomain 4.15.14-300.fc27.x86_64 #1 SMP Thu Mar 29 16:13:44 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
01:00.0 VGA compatible controller: NVIDIA Corporation GP106 [GeForce GTX 1060 6GB] (rev a1)
Hello,
I have encountered following error during installation DKMS kernel module:
-> Installing DKMS kernel module:
ERROR: Failed to run `/usr/sbin/dkms build -m nvidia -v 390.48 -k 4.13.9-300.fc27.x86_64`: Error! echo
Your kernel headers for kernel 4.13.9-300.fc27.x86_64 cannot be found at
/lib/modules/4.13.9-300.fc27.x86_64/build or /lib/modules/4.13.9-300.fc27.x86_64/source.
How can I fix this?
Hi Simaus,
You are running very old kernel?
Are you sure that you updated your system and reboot on (step 2.4 Make sure that you system is up-to-date and you are running latest kernel)?
Hi JR,
Huh yes, it’s old kernel. xd I’ve updated
Fedora (64-bit) and proper files have been installed when it was required but now
I have to update GRUB entries, because my PC has ‘a multiboot’ (the last installed system is Xubuntu 64-bit, but there is also Windows XP SP3 32-bit).
I will try to do it on my own when I’ll have time, but if you have any resolution how to update it in the best way I’ll try that.
Ok, so… it was only an old kernel issue. Here’s my output:
[[email protected] ~]$ nvidia-installer -v |grep version
nvidia-installer: version 390.48 ([email protected])
[[email protected] ~]$ uname -a
Linux libertania 4.15.14-300.fc27.x86_64 #1 SMP Thu Mar 29 16:13:44 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 620 OEM] (rev a1)
And here are two screenshots:
https://i.imgur.com/eZ6NLwb.png
https://i.imgur.com/yZr4fbX.png
Little display, but “/home sweet /home”. :P
Excellent! Nice to hear that you got it working!
[[email protected] ~]$ nvidia-installer -v |grep version
nvidia-installer: version 390.48 ([email protected]) Thu Mar 22 01:07:44 PDT 2018
[[email protected] ~]$ uname -a
Linux M4800Fedora 4.15.14-300.fc27.x86_64 #1 SMP Thu Mar 29 16:13:44 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 GK106GLM [Quadro K2100M] (rev a1)
[[email protected] ~]$
This is on a Dell M4800 laptop (K2100M 3200×1800) which natively supports three external monitors:
-HDMI: Dell 32″ 3840×2160
-DisplayPort: LG 3440×1440
-VGA: iiyama 2048×1152
Full desktp capture (13MB):
http://mega.nz/#!3dw3SSDZ!hl8HwZm9e_x5yevlQeajhJ3fMPoTglLOsQAOr03N0FM
Windows 10 and Fedora 27 installed on separate disks (F12 on boot to choose OS); Fedora 27 on a 500 GB internal mSATA. Fedora 27 was installed without the three external monitors connected, since with the installation display driver the installation hangs.
Following this guide nVidia driver was smoothly installed, and the three external monitors were then connected. Everything works beautifully.
Excellent! Nice to hear that you got everything working!
After some updates and after upgrading to Fedora 28, I had to repeat driver installation, with “export IGNORE_CC_MISMATCH=1”.
[email protected] ~]$ nvidia-installer -v |grep version
nvidia-installer: version 396.24 ([email protected])
[[email protected] ~]$ uname -a
Linux M4800Fedora 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
[[email protected] ~]$ lspci |grep -E “VGA|3D”
01:00.0 VGA compatible controller: NVIDIA Corporation GK106GLM [Quadro K2100M] (rev a1)
Laptop screen and the three external monitors work as before.
nvidia-installer: version 384.130 ([email protected]) Wed Mar 21 04:09:15 PDT 2018
Linux my-fedora 4.15.14-300.fc27.x86_64 #1 SMP Thu Mar 29 16:13:44 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
01:00.0 VGA compatible controller: NVIDIA Corporation GP104 [GeForce GTX 1080] (rev a1)
Many thanks for the guide!
# nvidia-installer -v |grep version
nvidia-installer: version 390.48 ([email protected])
# uname -a
Linux localhost.localdomain 4.15.15-300.fc27.x86_64 #1 SMP Mon Apr 2 23:14:02 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
# lspci |grep -E “VGA|3D”
01:00.0 VGA compatible controller: NVIDIA Corporation GP104M [GeForce GTX 1070 Mobile] (rev a1)
Thank you very much for this guide. Though, sometimes after a kernel update Gnome crashes and I have to run the installer again (in runlevel 3) to get it working.
Is this normal ?
(See my comment # 318983 at https://www.if-not-true-then-false.com/2015/fedora-nvidia-guide/comment-page-60/#comment-318983).
Thank you!
Andrea
Hi Andrea,
This isn’t normal, DKMS should take care updates, but unfortunately older nVidia driver versions (304.xx and 340.xx) sometimes do this. Does this same happen, if you update just kernel packages and nothing else?
Hi JR,
thank you for your reply. I suppose that the word “sometimes” in my previous comment (and also in yours) could be related to your question. I will pay attention to it on next update and report here. Thanks!
Andrea
Upgraded only kernel packages (4.15.15-300 –> 4.15.16-300), and no problems.
So I suppose that, when I have some updates to be installed, I can workaround the problem by installing at first the kernel packages only, then the remaining ones.
But this wouldn’t be a definitive, clean solution.
So, is there anything useful I can do to investigate further? Is it more likely a driver or dkms problem, or both together, or something else? Would it be worth it to file a bug (where?)?
Should I provide more info here to get some advice?
Thank you very much!
Andrea
Hi Andrea,
Actually this is not dkms problem, dkms works as it should and you can update kernel with other packages, but there is some other updates, example xorg or mesa, which override nVidia drivers files (then you have to rerun installer get this nVidia drivers files back).
And this problem concerns only nVidia 304.xx and 340.xx drivers, later versions works without this problem. Unfortunately only nVidia can fix this, but they doesn’t do active development with older drivers. Of course you can do some hacking with own scripts and replace updated files back, but running installer on runlevel 3 might be easier solution.
[[email protected] ~]# uname -a
Linux choquet.fritz.box 4.15.15-300.fc27.x86_64 #1 SMP Mon Apr 2 23:14:02 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
[[email protected] ~]# lspci |grep -E VGA
01:00.0 VGA compatible controller: NVIDIA Corporation G86 [GeForce 8500 GT] (rev a1)
Running NVIDIA-Linux-x86_64-340.106.run does not work for me:
[[email protected] ~]# tail -20 /var/lib/dkms/nvidia/340.106/build/make.log
^~~~~~~~~~~~~~~~~~~
/var/lib/dkms/nvidia/340.106/build/uvm/nvidia_uvm_lite.c: Auf höchster Ebene:
/var/lib/dkms/nvidia/340.106/build/uvm/nvidia_uvm_lite.c:857:14: Fehler: Initialisierung von inkompatiblem Zeigertyp [-Werror=incompatible-pointer-types]
.fault = _fault,
^~~~~~
/var/lib/dkms/nvidia/340.106/build/uvm/nvidia_uvm_lite.c:857:14: Anmerkung: (nahe der Initialisierung für »uvmlite_vma_ops.fault«)
/var/lib/dkms/nvidia/340.106/build/uvm/nvidia_uvm_lite.c:887:14: Fehler: Initialisierung von inkompatiblem Zeigertyp [-Werror=incompatible-pointer-types]
.fault = _sigbus_fault,
^~~~~~~~~~~~~
/var/lib/dkms/nvidia/340.106/build/uvm/nvidia_uvm_lite.c:887:14: Anmerkung: (nahe der Initialisierung für »counters_vma_ops.fault«)
cc1: Einige Warnungen werden als Fehler behandelt
make[3]: *** [/usr/src/kernels/4.15.15-300.fc27.x86_64/scripts/Makefile.build:325: /var/lib/dkms/nvidia/340.106/build/uvm/nvidia_uvm_lite.o] Fehler 1
make[2]: *** [/usr/src/kernels/4.15.15-300.fc27.x86_64/Makefile:1527: _module_/var/lib/dkms/nvidia/340.106/build/uvm] Fehler 2
make[2]: Verzeichnis „/usr/src/kernels/4.15.15-300.fc27.x86_64“ wird verlassen
make[1]: *** [Makefile:146: sub-make] Fehler 2
make[1]: Verzeichnis „/usr/src/kernels/4.15.15-300.fc27.x86_64“ wird verlassen
NVIDIA: left KBUILD.
nvidia-uvm.ko failed to build!
make: *** [Makefile:216: nvidia-uvm.ko] Fehler 1
make: Verzeichnis „/var/lib/dkms/nvidia/340.106/build/uvm“ wird verlassen
Downloading the patched 340.106 archive from your site does not work…
Hi Andreas,
Did you really tried to install patched version of 340.106 drivers? I just tested patched version with 4.15.15 kernel and it build, installs and works without errors.
is there any guide on how to update the installed driver? Been googling for a while, and coming up short
Hi belku,
Yes, you can update nVidia drivers running following steps again:
2.1 Download nVidia Installer Package
2.2 Make nVidia installer executable
2.7 Reboot to runlevel 3
2.8 Install nVidia proprietary drivers for GeForce 6/7 & GeForce 8/9/200/300 & GeForce 400/500/600/700/800/900/10 series cards (all installer steps)
2.9 All Is Done and Then Reboot Back to Runlevel 5
Hi there
Worked like a charm, thx a lot.
Now, in init 3, it’s good to know that you can switch between terminals with ‘alt+Fx’ combination if needed.
[[email protected] ~]$ nvidia-installer -v |grep version
nvidia-installer: version 390.48 ([email protected]) Thu Mar 22 01:07:44 PDT
[[email protected] ~]$ uname -a
Linux linux.home 4.15.15-300.fc27.x86_64 #1 SMP Mon Apr 2 23:14:02 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
[paul[email protected] ~]$ lspci |grep -E “VGA|3D”
01:00.0 VGA compatible controller: NVIDIA Corporation GK208B [GeForce GT 710] (rev a1)
[[email protected] ~]$
nvidia-installer: version 384.130 ([email protected])
Linux mtech96.localdomain 4.15.16-300.fc27.x86_64 #1 SMP Mon Apr 9 17:50:06 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
00:02.0 VGA compatible controller: Intel Corporation HD Graphics 620 (rev 02)
01:00.0 3D controller: NVIDIA Corporation GM108M [GeForce 940MX] (rev a2)
I followed all the instructions on a spanking fresh Fedora 27 install on a Dell 15 XPS 9560.
During the install, I DID NOT GET THE MESSAGE:
“Would you like to run the NVIDIA X-CONFIG utility…” (step 2.8.7).
It just said, “Done”
On reboot, my system is using the Intel Graphics…
[[email protected] ~]$ nvidia-installer -v |grep version
nvidia-installer: version 390.48 ([email protected])
[[email protected] ~]$ uname -a
Linux FedoraXPS 4.15.16-300.fc27.x86_64 #1 SMP Mon Apr 9 17:50:06 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
[[email protected] ~]$ lspci |grep -E “VGA|3D”
00:02.0 VGA compatible controller: Intel Corporation Device 591b (rev 04)
01:00.0 3D controller: NVIDIA Corporation GP107M [GeForce GTX 1050 Mobile] (rev a1)
When I look at the details box, it says “Intel Graphics”
How can I fix my X-config file to use NVidia?
Hi Heinrich,
I guess that problem is not your xorg.conf file, it’s your nVidia Optimus device (step 1.2). nVidia support for Optimus devices on Linux is limited and you can only do offscreen rendering and run CUDA apps, but can’t drive a display. More info from http://download.nvidia.com/XFree86/Linux-x86_64/390.48/README/optimus.html
You can of course try https://bumblebee-project.org/ it should work with nVidia Optimus devices, but I have read some discussions about poor performance.
Hello,
Thanks for the guide! I have run into some issues though:
I’ve got a fresh install of Fedora 27, have followed the guide to the letter, no errors, the installation went OK. Setup I’ve got is 2 NVS 510 cards, 8 monitors, 4 hooked up to each card. After installing the driver, 4 monitors are on and displaying the desktop OK. The other 4, however, are displaying black picture. After moving the mouse into the screen area of any of the 4 screens displaying black picture, the mouse cursor is visible, but changes to an X.
I have since removed the drivers as I could not get all 8 screens to display the desktop.
Have you come across this problem and if so can you suggest a fix, please?
lspci |grep -E “VGA|3D”
03:00.0 VGA compatible controller: NVIDIA Corporation GK107 [NVS 510] (rev a1)
04:00.0 VGA compatible controller: NVIDIA Corporation GK107 [NVS 510] (rev a1)
uname -a
Linux screens 4.15.16-300.fc27.x86_64 #1 SMP Mon Apr 9 17:50:06 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
Kernel has been updated to the latest version and I have tried the following drivers (each with fresh OS install), but ended up with the 4 screens unusable each time.
ls nvdrv/
NVIDIA-Linux-x86_64-384.130.run NVIDIA-Linux-x86_64-384.98.run NVIDIA-Linux-x86_64-390.48.run NVIDIA-Linux-x86_64-396.18.run
Thanks
Hello tomski,
Sounds very nice setup. Have to say, that I have never seen setup like this. :)
First few questions. What desktop you are using? And do you have tested this same setup with some other OS, like Windows is it working normally?
Then could you post output of following command:
Howdy!
Thanks for the reply! I finally (well not me myself but a friend of mine) figured it out. In nvidia-setttings panel, after enabling (disabled by default) the other 4 monitors (the ones I said were displaying black picture), you have to attach them to “screen 0”, or the same “screen X” the first 4, which were working OK, were attached.
I, wrongly, assumed that the first card was referred to as “screen 0” in the nvidia-settings panel and the second one was “screen 1”, so what I have been doing all that time was, instead of attaching these other 4 monitors to “screen 0” where working ones were attached, I was assigning them to (new) “screen 1”, which was, misleadingly, the default option from the drop down list when enabling new monitors. Oh, and of course I could not see the option to assign them to “screen 0” anywhere #facepalm. Another PEBCAK example.
Anyway, it’s working flawlessly now!
To answer your question – the pc is Dell precision tower 7910. We use it at work to display pretty graphs and other things ;)
Oh and btw, driver verios used is 390.48 and kernel version is 4.15.17-300.fc27.x86_64 #1
Working UI is GDM v. 3.26.2.1
Again, thank you LOTS for your awesome guide!
T
Well thanks a lot for a comprehensive guide, it totally made my day enabling all that graphical niceness in KDE.
However I’ve ran in to that nasty issue with Plymouth rejecting graphical themes on boot. Only showing installed theme on shutdown but always falling back to the default tri-block low-res theme on startup. Only text-based ‘Breeze-text’ theme with basic progress bar worked.
I’ve noticed these lines in Plymouth debug log:
[ply-device-manager.c:274] create_devices_for_subsystem:creating objects for frame buffer devices
[ply-device-manager.c:291] create_devices_for_subsystem:found device /sys/devices/platform/vesa-framebuffer.0/graphics/fb0
[ply-device-manager.c:315] create_devices_for_subsystem:it’s not initialized
[ply-device-manager.c:291] create_devices_for_subsystem:found device /sys/devices/virtual/graphics/fbcon
[ply-device-manager.c:315] create_devices_for_subsystem:it’s not initialized
It’s a BIOS based system with GeForce GTX 660 card.
Any help on beating this one would be appreciated.
Hello,
Looks like, I have to go through this Plymouth part again. In theory installation should be easy, but sometimes proprietary drivers won’t work so well with Fedora.
Thanks a lot. Really nice guide. Worked perfectly.
[[email protected] xakaitetoia]# nvidia-installer -v |grep version
nvidia-installer: version 396.18 ([email protected])
[[email protected] xakaitetoia]# uname -a
Linux xak-fedora 4.15.17-300.fc27.x86_64 #1 SMP Thu Apr 12 18:19:17 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
[[email protected] xakaitetoia]#
[[email protected] xakaitetoia]# lspci |grep -E “VGA|3D”
01:00.0 VGA compatible controller: NVIDIA Corporation GP107 [GeForce GTX 1050 Ti] (rev a1)