I have an NVIDIA GTX970 card, however, but I keep a GTX770 around because I know the combo of Kernel 9.10 and following doesn’t support the newer closed source drivers out of the box. I use Leap 42.2 with the NVIDIA driver there. But, will there come a time when the newer NVIDIA cards be supported in TW “out of the box” so I don’t have to fiddle with re-compling, etc? Or, is it better to switch to a newer AMD card and use Mesa? Can MESA be used with the newer AMD cards without compiling new driver and blacklisting Nouveau? Thanks for everyone’s input!!!
I’d guess never TW changes too fast for NVIDIA to keep up. Use Leap if you don’t want to fuss with it. In any case propritary drivers will never come in the same box as openSUSE. Note the name
MESA is part of the X stack. It is not a driver the new AMD drivers are AMDGPU and AMDGPU-PRO. But they are very very new and seem to be a bit shaky (works for some others not so much)
It might be possible to avoid recompiling on TW, but you’ll still face issues with major kernel updates where NVIDIA hasn’t had time to fix their drivers yet.
You may find it easier to handle by using an LTS kernel version, but standard disclaimers about stability would apply here.
It’s unlikely that Nouveau will ever fully support the newer cards due to changes made by NVIDIA starting from Maxwell.
Mesa is the part of the default open source graphics stack, so for newer AMD cards there is no reason to have to compile new drivers or blacklist anything in configuration files.
Thankyou.SoundslikeAMDisthewaytogo!
Sound like AMD is the way to go. Thanks. I’ll have to stay with the 4.4 kernel for now anyway. Thanks again!!!
I use the proprietory nvidia driver in both TW and leap. If you put the instructions in a short bash script, they take a few seconds to compile when the kernel changes.
In TW I like to try the new kernels so the drivers needed patching for 4.10 and 4.11-rc. But even that’s not so difficult and you only need to patch them once for each new kernel version.
Hey there fellow proprietary nvidia driver user.
Would you mind doing two things?
- post the contents of your short bash script
- tell a fellow (me) where you got your patches from 4.10 and 4.11-rc
I had already posted for help on getting the patches to apply successfully but no one replied.
Advice I searched out myself hinted that the patch for 4.9 ought to be applied along with the 4.10 patch. The 4.9 patch failed with a message that the patch was garbage.
I got the patches from the nvidia forums so I was surprised to read of the success of at least one other user who applied both.
There were also links in our own forums but they tended to lead back to the same two patches.
My nvidia card is the 340.xx series
Apologies to the thread starter.
Your help is appreciated,
H
I keep the nvidia downloads in /home/updates/ and the patched file in /home updates/411/. Obviously the 411 could change to 409, 410 or perhaps, 412 depending on the kernel. If you change that then you would need to change the script approriately. You run the script as root (su or sudo) after loading new kernel to level 3 (no X).
#!/bin/bash
if -n $1 ]; then
echo "Do you want standard NVIDIA?(y/411) :"
read myanswer
if $myanswer = "y" ]; then
dir="/home/updates/"
else
dir="/home/updates/nv-$myanswer"
fi
fi
if -e /usr/bin/nvidia-uninstall ]; then
nvidia-uninstall
fi
cd $dir
sh NVIDIA*run -a --no-cc-version-check --install-libglvnd --glvnd-glx-client -q
exit
The patches can be applied by either extracting the nvidia driver file and applying the patch appropriately or by using
sh NVIDIA*run --apply-patch /path/to/patch
from folder containing driver.
If you get a query as to which file to patch it means the patch can’t find the level of the file. It will quote the full path within the driver file. You just need to tell it the file. You can copy it from kernel/… in the error msg and paste it in the query (drop leading"/" before kernel/…). There may be 4 or 5 of these. Just do it for each new file query. It sounds complicated but it just takes seconds.
There’s probably a “p” option (check ‘man patch’ in a terminal) to tell the apply-patch command but I haven’t been able to find it.
If you need any further explanations of my attempt to explain, let me know here.
Ps I get the patches here: (https://devtalk.nvidia.com/default/topic/1002820/linux/-patch-381-09-kernel-4-11-rc5/)
[QUOTE=peteh100;2819418]I keep the nvidia downloads in /home/updates/ and the patched file in /home updates/411/. Obviously the 411 could change to 409, 410 or perhaps, 412 depending on the kernel. If you change that then you would need to change the script approriately. You run the script as root (su or sudo) after loading new kernel to level 3 (no X).
#!/bin/bash
if -n $1 ]; then
echo "Do you want standard NVIDIA?(y/411) :"
read myanswer
if $myanswer = "y" ]; then
dir="/home/updates/"
else
dir="/home/updates/nv-$myanswer"
fi
fi
if -e /usr/bin/nvidia-uninstall ]; then
nvidia-uninstall
fi
cd $dir
sh NVIDIA*run -a --no-cc-version-check --install-libglvnd --glvnd-glx-client -q
exit
The patches can be applied by either extracting the nvidia driver file and applying the patch appropriately or by using
sh NVIDIA*run --apply-patch /path/to/patch
from folder containing driver.
If you get a query as to which file to patch it means the patch can’t find the level of the file. It will quote the full path within the driver file. You just need to tell it the file. You can copy it from kernel/… in the error msg and paste it in the query (drop leading"/" before kernel/…). There may be 4 or 5 of these. Just do it for each new file query. It sounds complicated but it just takes seconds.
There’s probably a “p” option (check ‘man patch’ in a terminal) to tell the apply-patch command but I haven’t been able to find it.
If you need any further explanations of my attempt to explain, let me know here.
Ps I get the patches here: [
Pete,
Thanks for the reply and the script.
I thought I ought to clear something here.
I don’t have any difficulty applying patches to kernels or dropping to runlevel 3.
That’s not the issue. The patches for 4.10 and 4.9 simply would not apply to the 340.102 nvidia run driver.
I notice that the patch you are pointing to is for the 381 driver.
I’ll give it a shot but I’m beginning to think that I’m not going to see a patch for my 340.102.
I’ll check the nvidia devtalk threads again for my specific driver.
In the meantime,
cheers]("http://[https://devtalk.nvidia.com/default/topic/1002820/linux/-patch-381-09-kernel-4-11-rc5)
Yes, sorry. I was really just pointing to the site. I’ve found all my recent patches there.
Perhaps if you search within that you’ll come up with something.
No worries.
I’ve checked the site again but it’s still the same stuff from the last time.
Thanks for your time and input,
H
Hi again Pete,
There is something else you can help me with - if you don’t mind.
When you get to the part of the nvidia installation where it asks if you want to set DKMS, the default is No.
I’d already pre-installed the dkms package from Packman - and decided to enter Yes.
The installer failed straight after that.
After recovering rebooting etc. I re-ran the installer let it unistall the remants of the failed install ( I know, I know…it’s a thankless task when doing it the hard way) and rebooted once again.
This time I stayed with the default NO at the DKMS prompt and the install proceed until it failed with something else - I don’t need to talk about that part just yet.
What did you choose to do at the DKMS part of the nvidia install? NO/YES?
cheers,
Hugh
Hi Pete,
Please ignore the previous requests.
I’ve done with it - no more!
Back to Leap for me.
Thanks once again for your inputs,
Hugh