I have set up nvidia driver / cuda using deb and runfile installation methods on more than a dozen different computers all running ubuntu 16.04 - the desktop installs typically are very stable but the server editions have a 50/50 chance of the driver failing to build against a kernel upgrade - I’ve tried mixing up which version of the driver I install but it doesn’t seem to make a difference. Why is it so unstable on the server edition and is there anything I can do to improve stability? I find it particularly confusing as I thought that the server and desktop kernels were the same?
Related topics
Topic | Replies | Views | Activity | |
---|---|---|---|---|
CUDA installation in Ubuntu server Edition | 0 | 812 | August 12, 2015 | |
new stable kernel 3.9.7 and nvidia-settings - how to solve ? | 0 | 816 | June 22, 2013 | |
Ubuntu 16.04 CUDA 8+driver install problems with mixed sources | 2 | 2463 | March 30, 2017 | |
Failed Cuda Driver and Runtime version may be mismatched Cuda installation fails on Ubuntu 10.4 x86_ | 13 | 5090 | November 17, 2010 | |
Why we still has to handle the `Failed to initialize NVML: Driver/library version mismatch` in 2024? | 0 | 272 | June 11, 2024 | |
Best Linux distro for CUDA development | 3 | 5644 | March 3, 2016 | |
Mismatch in CUDA driver and runtime versions | 6 | 1090 | September 3, 2024 | |
cuda installation compatibility | 2 | 608 | June 21, 2017 | |
Back to a non Cuda driver | 2 | 1823 | January 13, 2009 | |
Unable to setup Cuda and Nvidia drivers | 0 | 219 | June 8, 2024 |