Open HarshaMupparaju opened 1 year ago
Hi Harsha,
I never bumped into this issue myself, but I did find this hack that seems to have helped people with the same issue - perhaps give that a try? https://github.com/NVIDIA/nccl/issues/650#issuecomment-1145173577
Best, Jamie
i have met this problem in DVGO, maybe upgrad your g++ or gcc can be helpful
I have met the same problem, do you solve it? I have a lot of questions on solving the environment on my own machine because the cuda version or g++ version is not matching the author's, so I run it on the AutoDL and it's simple to configure the environment. Detail: cuda 11.6 , Ubuntu 20.04, python 3.8. You can remote on your own machine via vscode or pycharm. I wish it could help.
I have cloned diffusionerf and tried to execute
conda env create -f environment.yml
Commented out the line
- git+https://github.com/NVlabs/tiny-cuda-nn/#subdirectory=bindings/torch
from environment.yml file, and tried creating the environment, it worked.Tried to pip install
git+https://github.com/NVlabs/tiny-cuda-nn/#subdirectory=bindings/torch
seperately and got this error:System settings: Ubuntu 22.04, gcc 11.4
Output of
nvidia-smi
:Output of nvcc --version:
Any help to resolve this issue?