Question about the latest NVidia proprietary driver update. Running Ubuntu Studio 24.04.4 LTS 6.8.0-136 lowlatency. NVidia 535.309.1 on GeForce GTX 750 hardware, X11 windowing.
This morning’s update in Discover shows many libnvidia packages to be updated, as well as what appear to be packages to transition to the 580 driver. Also appears that many 535 libnvidia packages will be removed.
Also as of this morning the ubuntu-drivers devices command shows 580 as recommended. The last time I checked this a few days ago, 535 was still the recommended driver.
Something similar to this happened a few months back. The 580 driver showed as recommended for just a few hours. Updating caused significant issues for me, and I rolled back to 535 manually. Then less than a day later the 535 started showing as recommended again, and it’s been that way ever since up until today.
I’m very reluctant to allow this update to occur. Isn’t this what LTS releases are supposed to avoid? Please advise, especially if you have experience with a situation similar to mine.
Got it, @rubi1200 , thank you. Keeping a close eye on this one. Going back to my days waging battles with the ATI fglrx driver in Lenovo laptops, I have learned the hard way to use extreme caution any time a proprietary driver is involved.
I had problems with sleep mode. The 535 driver has always had problems waking up from sleep; the system would wake up with a blank display and be unresponsive to anything other than holding down the power button until it shut off. Fortunately this could be (and in fact still is, on my system) overcome by bypassing the script in /usr/bin/nvidia-sleep.sh. As I expected, this problem persisted when the 580 driver was (briefly) installed. However that bypass fix didn’t work. It just prevented the system from successfully sleeping; it would drop into sleep mode for two seconds then immediately wake up.
Thanks much for the comment, appreciate any further suggestions you (or anyone reading this) might have.
The suggestion, as always for this kind of issues, is to update firmware (UEFI) and troubleshoot from there. This seldom is a problem of OS/drivers (but can be corrected by software in some circumstances).
Of note, in your own words, no difference between drivers branches (and likely 580 had better performance), other than not being able to apply the script you’ve been using. So, 580 IS correct.
Thanks, sorry if my post was unclear. The /usr/bin/nvidia-sleep.sh isn’t mine, it comes from the driver installation. I just added “exit 0” as the first non-comment line. With the 535 driver this is all that’s needed to get the system sleeping and waking properly.
As for UEFI/firmware, I can see the thinking behind this comment but my computer is old enough (2014-vintage Lenovo desktop) that it would be one thing that would scare me more than the NVidia driver update.
Lenovo is good with that so it’s generally safe to update the firmware and there’s always a reason why manufacturers release new versions. As long as you follow their instructions it should be fine.
Yes. Not particularly scientifically, but yes. Every time a minor update to the 535 would come along and re-install the nvidia-sleep.sh, the failure to wake properly from sleep would reoccur. Telling the script to exit as soon as it was called reliably fixed it.
Since this was rather infrequent I opted to continue with the manual script hack, distasteful as it is, rather than trying to adjust the package update procedure. But very much appreciate your suggestion as it may come in handy in the near future
