[Bug] Unable to install MMCV 2.1.0 The detected CUDA version (11.8) mismatches the version that was used to compile PyTorch (12.1). Please make sure to use the same CUDA versions.

[Fernando961226]

Prerequisite

• I have searched Issues and Discussions but cannot get the expected help.
• The bug has not been fixed in the latest version(https://github.com/open-mmlab/mmcv).

Environment

• pip install torch torchvision torchaudio
• python -c 'import torch;print(torch.version)'
  2.3.0+cu121
• nvidia-smi
  Thu May 16 13:52:34 2024
  +---------------------------------------------------------------------------------------+
  | NVIDIA-SMI 535.161.08 Driver Version: 535.161.08 CUDA Version: 12.2 |
  |-----------------------------------------+----------------------+----------------------+

Reproduces the problem - code sample

N/A

Reproduces the problem - command or script

min install mmcv=2.1.0

Reproduces the problem - error message

Looking in links: https://download.openmmlab.com/mmcv/dist/cu121/torch2.3.0/index.html
Collecting mmcv==2.1.0
Using cached mmcv-2.1.0.tar.gz (471 kB)
Preparing metadata (setup.py) ... done
Requirement already satisfied: addict in /home/fernando/anaconda3/envs/sea-ice-fm/lib/python3.11/site-packages (from mmcv==2.1.0) (2.4.0)
Collecting mmengine>=0.3.0 (from mmcv==2.1.0)
Using cached mmengine-0.10.4-py3-none-any.whl.metadata (20 kB)
File "/home/fernando/anaconda3/envs/sea-ice-fm/lib/python3.11/site-packages/torch/utils/cpp_extension.py", line 522, in build_extensions .....

    File "/home/fernando/anaconda3/envs/sea-ice-fm/lib/python3.11/site-packages/torch/utils/cpp_extension.py", line 522, in build_extensions
      _check_cuda_version(compiler_name, compiler_version)
    File "/home/fernando/anaconda3/envs/sea-ice-fm/lib/python3.11/site-packages/torch/utils/cpp_extension.py", line 417, in _check_cuda_version
      raise RuntimeError(CUDA_MISMATCH_MESSAGE.format(cuda_str_version, torch.version.cuda))
  RuntimeError:
  The detected CUDA version (11.8) mismatches the version that was used to compile
  PyTorch (12.1). Please make sure to use the same CUDA versions.

Additional information

No response

[zhouzaida]

Hi, support for torch2.3.0 has only been available since mmcv2.2.0, you can install mmcv2.2.0 to avoid source compilation problems!

[fengduduaaa]

Hi, support for torch2.3.0 has only been available since mmcv2.2.0, you can install mmcv2.2.0 to avoid source compilation problems!

new bugs:"mmcv==2.2.0 is used but incompatible. Please install mmcv-full>=[2, 0, 0], <[2, 1, 0]."

[smallrookie]

Hi, support for torch2.3.0 has only been available since mmcv2.2.0, you can install mmcv2.2.0 to avoid source compilation problems!

new bugs:"mmcv==2.2.0 is used but incompatible. Please install mmcv-full>=[2, 0, 0], <[2, 1, 0]."

I had the same problem

[gg22mm]

pip install torch==2.1.2