Anybody here！ My GPU is 3080ti，BUt I have many promblems.Help

[LanqingGuo]

[LanqingGuo]

(edter) F:\Code\EDTER>pip install mmcv-full -f https://download.openmmlab.com/mmcv/dist/cu101/torch1.6.0/index.html
Looking in links: https://download.openmmlab.com/mmcv/dist/cu101/torch1.6.0/index.html
Collecting mmcv-full
Using cached mmcv-full-1.7.2.tar.gz (607 kB)
Preparing metadata (setup.py) ... done
Requirement already satisfied: addict in d:\program\anaconda3\envs\edter\lib\site-packages (from mmcv-full) (2.4.0)
Requirement already satisfied: numpy in d:\program\anaconda3\envs\edter\lib\site-packages (from mmcv-full) (1.21.5)
Collecting packaging
Using cached packaging-24.0-py3-none-any.whl (53 kB)
Requirement already satisfied: Pillow in d:\program\anaconda3\envs\edter\lib\site-packages (from mmcv-full) (9.4.0)
Requirement already satisfied: pyyaml in d:\program\anaconda3\envs\edter\lib\site-packages (from mmcv-full) (6.0.1)
Requirement already satisfied: yapf in d:\program\anaconda3\envs\edter\lib\site-packages (from mmcv-full) (0.40.2)
Requirement already satisfied: regex in d:\program\anaconda3\envs\edter\lib\site-packages (from mmcv-full) (2024.4.16)
Requirement already satisfied: platformdirs>=3.5.1 in d:\program\anaconda3\envs\edter\lib\site-packages (from yapf->mmcv-full) (4.0.0)
Requirement already satisfied: importlib-metadata>=6.6.0 in d:\program\anaconda3\envs\edter\lib\site-packages (from yapf->mmcv-full) (6.7.0)
Requirement already satisfied: tomli>=2.0.1 in d:\program\anaconda3\envs\edter\lib\site-packages (from yapf->mmcv-full) (2.0.1)
Requirement already satisfied: zipp>=0.5 in d:\program\anaconda3\envs\edter\lib\site-packages (from importlib-metadata>=6.6.0->yapf->mmcv-full) (3.15.0)
Requirement already satisfied: typing-extensions>=3.6.4 in d:\program\anaconda3\envs\edter\lib\site-packages (from importlib-metadata>=6.6.0->yapf->mmcv-full) (4.7.1)
Building wheels for collected packages: mmcv-full
Building wheel for mmcv-full (setup.py) ... error
error: subprocess-exited-with-error

× python setup.py bdist_wheel did not run successfully.
│ exit code: 1
╰─> [1630 lines of output]
C:\Users\hpuau\AppData\Local\Temp\pip-install-h0g55j97\mmcv-full_ac25281e1ae742c490f06ca385bc8ad6\setup.py:6: DeprecationWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html
from pkg_resources import DistributionNotFound, get_distribution, parse_version
running bdist_wheel
running build
running build_py
creating build
creating build\lib.win-amd64-cpython-37
creating build\lib.win-amd64-cpython-37\mmcv
copying mmcv\version.py -> build\lib.win-amd64-cpython-37\mmcv
copying mmcv_init_.py -> build\lib.win-amd64-cpython-37\mmcv
creating build\lib.win-amd64-cpython-37\mmcv\arraymisc

[LanqingGuo]

× python setup.py bdist_wheel did not run successfully.
│ exit code: 1
╰─> [1630 lines of output]
C:\Users\hpuau\AppData\Local\Temp\pip-install-h0g55j97\mmcv-full_ac25281e1ae742c490f06ca385bc8ad6\setup.py:6: DeprecationWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html
from pkg_resources import DistributionNotFound, get_distribution, parse_version

why exist this promblem

[LanqingGuo]

我要进行测试，按照提示进行环境准备，在安装mmc那行代码依赖时，提示如上，难道mmcv已经不支持3080ti了？

[leserein17]

I’m honestly shocked by how unreasonably strict this project is with version requirements. When I tried to follow the instructions in the readme.md on my 3090 GPU, I encountered the error:
RuntimeError: CUDA error: no kernel image is available for execution on the device,
which likely means the compute capability is too high.

So I attempted to downgrade to a more compatible setup—Torch 1.7.1 with CUDA 11.0, and updated mmcv-full to 1.7.1 accordingly. However, the project failed the internal check: MMCV_MIN = '1.1.2' MMCV_MAX = '1.2.7'. When I bypassed this version constraint, I ran into another error:
ImportError: libcudart.so.10.1: cannot open shared object file: No such file or directory
—clearly the project is trying to drag me back to CUDA 10.1.

I can’t believe a project would lock itself to such outdated and fragile versions, especially with this level of inflexibility.