Skip to content

Problem with installation and CUDA Version #16

@plpalmer

Description

@plpalmer

Hi,

I am facing some difficulty setting up the library, especially while utilizing different Cuda versions (10 and 11)

The First step of the setup works fine with recommended PyTorch 1.4 on Cuda 10.0

But for building OpenPCDet with Cuda10 I am getting the Error:
ValueError: Unknown CUDA arch (8.6) or GPU not supported

which seems to be related to my graphics card which is to new: https://stackoverflow.com/questions/66968382/valueerror-unknown-cuda-arch-8-6-or-gpu-not-supported

With a newer PyTorch Version like 1.13 on Cuda 11.7 I can build OpenPCDet but can not execute the first setup script.
I am getting the error:
File "/home/usr/anaconda3/envs/pttr/lib/python3.7/site-packages/torch/utils/cpp_extension.py", line 1916, in _run_ninja_build raise RuntimeError(message) from e RuntimeError: Error compiling objects for extension

Does someone have the same issue or has some possible solution for this problem?

Thanks in advance

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions