installing nvidia-cuda-toolkit removes nvidia-driver-455

[Lishai Eitan]

There seems to be a conflict between (the dependencies of) nvidia-cuda-toolkit and nvidia-driver-455.
At least one dependency of nvidia-cuda-toolkit has sthe following dependencies:
Package: libnvidia-ml-dev
...
Depends: libnvidia-compute-450 (>= 450) | libnvidia-compute-450-server (>= 450) | libnvidia-ml.so.1

However nvidia-driver-455 ultimately depends on libnvidia-compute-455, which does not satisfy libnvidia-compute-450 (>= 450).

ProblemType: Bug
DistroRelease: Ubuntu 20.10
Package: nvidia-cuda-toolkit (not installed)
ProcVersionSignature: Ubuntu 5.8.0-25.26-generic 5.8.14
Uname: Linux 5.8.0-25-generic x86_64
NonfreeKernelModules: nvidia_modeset nvidia
ApportVersion: 2.20.11-0ubuntu50
Architecture: amd64
CasperMD5CheckResult: skip
CurrentDesktop: KDE
Date: Tue Oct 20 01:25:35 2020
InstallationDate: Installed on 2020-10-19 (0 days ago)
InstallationMedia: Kubuntu 20.10 "Groovy Gorilla" - Beta amd64 (20200930)
SourcePackage: nvidia-cuda-toolkit
UpgradeStatus: No upgrade log present (probably fresh install)

[Launchpad Janitor (janitor)]

This bug was fixed in the package nvidia-cuda-toolkit - 11.0.3-1ubuntu1

---------------
nvidia-cuda-toolkit (11.0.3-1ubuntu1) groovy; urgency=medium

  * Update libcuda1 availability for groovy (LP: #1900627)

 -- Graham Inggs <email address hidden> Tue, 20 Oct 2020 12:30:55 +0000

[Lishai Eitan (lishai-eitan)]

Comment Summary:
----------------
suggested fix: make libnvidia-ml-dev depend on:
libnvidia-compute-450 (>= 450) | libnvidia-compute-450-server (>= 450) | libnvidia-ml.so.1 (>= 450) | libnvidia-ml1 (>=450)
instead of:
libnvidia-compute-450 (>= 450) | libnvidia-compute-450-server (>= 450) | libnvidia-ml.so.1 (>= 450)

Longer version :)
-----------------
I noticed now that libnvidia-ml.so.1 (>= 450) would satisfy libnvidia-ml-dev's dependencies (and in turn, would allow the installation of nvidia-cuda-toolkit).

both libnvidia-compute-450 and libnvidia-compute-455 provide libnvidia-ml1, AND
 contain the file libnvidia-ml.so.1, but do not "provide" libnvidia-ml.so.1, which suggests the problem is only in the dependency *declerations*, and nothing "material" prevents the installation of these packages together (i.e. if we would force their installation, they would work).

If this is correct, I believe that changing libnvidia-ml-dev to depend on libnvidia-ml1 (>=450) as an alternative to the dependency on libnvidia-ml.so.1 would fix the problem.

To test this theory, I created a dummy package (using equivs) that contain no files, depends on libnvidia-ml1 (>= 455.28) and "provides" libnvidia-ml.so.1 (= 455.28). After installing this package, I successfully installed nvidia-cuda-toolkit along with nvidia-driver-455.
This setup works for me: I was able to install pytorch and train some models on the gpu (verified using nvidia-smi).

[Lishai Eitan (lishai-eitan)]

Sorry, must have missed the fix released...
Thank you for the fix :)