mamba fails installing cuda-nvcc and cudatoolkit together

[pitrou]

Troubleshooting docs

• [X] My problem is not solved in the Troubleshooting docs

How did you install Mamba?

Mambaforge

Search tried in issue tracker

cuda

Latest version of Mamba

• [X] My problem is not solved with the latest version

Tried in Conda?

I have this problem with Conda as well, without using Mamba

Describe your issue

There are conflicts, which look spurious, when I try to install both cuda-nvcc and cudatoolkit:

$ mamba install -c conda-forge cuda-nvcc cudatoolkit

Looking for: ['cuda-nvcc', 'cudatoolkit']

conda-forge/linux-64                                        Using cache
conda-forge/noarch                                          Using cache
nodefaults/noarch                                             No change
nodefaults/linux-64                                           No change

Pinned packages:
  - python 3.10.*

Could not solve for environment specs
The following packages are incompatible
├─ cuda-nvcc is installable and it requires
│  └─ cuda-nvcc_linux-64 12.0.76.* , which requires
│     └─ cuda-nvcc-dev_linux-64 12.0.76.* , which requires
│        └─ cuda-version >=12.0,<12.1.0a0 , which requires
│           └─ cudatoolkit 12.0|12.0.* , which can be installed;
└─ cudatoolkit is not installable because it conflicts with any installable versions previously reported.

mamba info / micromamba info

mamba version : 1.4.9
     active environment : pyarrow
    active env location : /home/antoine/mambaforge/envs/pyarrow
            shell level : 1
       user config file : /home/antoine/.condarc
 populated config files : /home/antoine/mambaforge/.condarc
                          /home/antoine/.condarc
          conda version : 23.5.2
    conda-build version : not installed
         python version : 3.9.16.final.0
       virtual packages : __archspec=1=x86_64
                          __cuda=12.0=0
                          __glibc=2.35=0
                          __linux=5.15.0=0
                          __unix=0=0
       base environment : /home/antoine/mambaforge  (writable)
      conda av data dir : /home/antoine/mambaforge/etc/conda
  conda av metadata url : None
           channel URLs : https://conda.anaconda.org/conda-forge/linux-64
                          https://conda.anaconda.org/conda-forge/noarch
                          https://conda.anaconda.org/nodefaults/linux-64
                          https://conda.anaconda.org/nodefaults/noarch
          package cache : /home/antoine/mambaforge/pkgs
                          /home/antoine/.conda/pkgs
       envs directories : /home/antoine/mambaforge/envs
                          /home/antoine/.conda/envs
               platform : linux-64
             user-agent : conda/23.5.2 requests/2.31.0 CPython/3.9.16 Linux/5.15.0-78-generic ubuntu/22.04.2 glibc/2.35
                UID:GID : 1000:1000
             netrc file : None
           offline mode : False

Logs

Posted on https://gist.github.com/pitrou/8c8a44346b073ab65f7dacaf061b1bc9 , since Github would otherwise complain ("body is too long (maximum is 65536 characters)").

environment.yml

No response

~/.condarc

channels:
  - conda-forge
  - nodefaults

auto_update_conda: false
show_channel_urls: true

auto_activate_base: false
channel_priority: strict
disallowed_packages: []

[pitrou]

As mentioned above, I had to post the logs separately as they were too long to fit: https://gist.github.com/pitrou/8c8a44346b073ab65f7dacaf061b1bc9

[jonashaag]

Looks like a bug because your system seems to have the right CUDA version installed and the solver error messages are conflicting

[pitrou]

So it looks like it's probably a conda-forge issue instead: https://github.com/conda-forge/staged-recipes/issues/21382

Why am I not surprised... sorry for this issue.