Error to install RAPIDS libraries into project

[lucaskbobadilla]

I am trying to install the accelarated pandas version from rapids but I am getting the following error:

• If I try to use: uv add cudf-cu12

I am getting the following error:

Building cudf-cu12==24.8.2
⠼ cudf-cu12==24.8.2                                                                                                                                                                                                                                                             error: Failed to download and build `cudf-cu12==24.8.2`
  Caused by: Build backend failed to build wheel through `build_wheel()` with exit status: 1
--- stdout:

--- stderr:

nvidia_stub.error.InstallFailedError:
*******************************************************************************

The installation of cudf-cu12 for version 24.8.2 failed.

This is a special placeholder package which downloads a real wheel package
from https://pypi.nvidia.com. If https://pypi.nvidia.com is not reachable, we
cannot download the real wheel file to install.

You might try installing this package via

$ pip install --extra-index-url https://pypi.nvidia.com cudf-cu12


Here is some debug information about your platform to include in any bug
report:

Python Version: CPython 3.12.5
Operating System: Linux 4.18.0-477.27.1.el8_8.x86_64
CPU Architecture: x86_64
nvidia-smi command not found. Ensure NVIDIA drivers are installed.

*******************************************************************************
---

• I tried to use pip too as suggested in the error: uv pip install --extra-index-url https://pypi.nvidia.com cudf-cu12

  × No solution found when resolving dependencies:
  ╰─▶ Because only the following versions of cudf-cu12 are available:
          cudf-cu12==23.6.0
          cudf-cu12==23.6.1
          cudf-cu12==23.8.0
          cudf-cu12==23.10.0
          cudf-cu12==23.10.1
          cudf-cu12==23.10.2
          cudf-cu12==23.12.0
          cudf-cu12==23.12.1
          cudf-cu12==24.2.0
          cudf-cu12==24.2.1
          cudf-cu12==24.2.2
          cudf-cu12==24.4.0
          cudf-cu12==24.4.1
          cudf-cu12==24.6.0
          cudf-cu12==24.6.1
          cudf-cu12==24.8.1
          cudf-cu12==24.8.2
      and all versions of cudf-cu12 have no wheels with a matching Python ABI tag, we can conclude that cudf-cu12<23.6.1 cannot be used.
      And because you require cudf-cu12, we can conclude that your requirements are unsatisfiable.

uv version: 0.40

[zanieb]

Hi! The error is from the package build (as well as the hint about using pip). I think the relevant line is:

nvidia-smi command not found. Ensure NVIDIA drivers are installed.

Do you have that command?

[lucaskbobadilla]

Yeah, I was not activating CUDA before. I activate it now to have nvidia-smi available but the error still persists. I also noticed some SSL error from uv requests. Is there a way to update the certificates uv can access?

[zanieb]

@lucaskbobadilla you can use --native-tls to use your system certificates.

[abc8747]

Python 3.12 is not yet officially supported in the stable channel - switching to the pre-release version worked for me: pyproject.toml

[project]
name = ""
version = ""
description = ""
readme = "README.md"
requires-python = ">=3.12"
dependencies = [
  "cudf-cu12"
]

[tool.uv]
extra-index-url = ["https://pypi.anaconda.org/rapidsai-wheels-nightly/simple"]
index-strategy = "unsafe-best-match"
prerelease = "allow"

[lucaskbobadilla]

@cathaypacific8747 That worked for me