conda install pyg -c pyg hangs on Solving environment: \

[moschles]

> conda install pyg -c pyg
Collecting package metadata (current_repodata.json): done
Solving environment: failed with initial frozen solve. Retrying with flexible solve.
Solving environment: failed with repodata from current_repodata.json, will retry with next repodata source.
Collecting package metadata (repodata.json): done
Solving environment: failed with initial frozen solve. Retrying with flexible solve.
Solving environment: \

This hangs here for hours and nothing happens.
.

Python 3.8.13
ogb-1.3.4-pyhd8ed1ab_0
pytorch-1.6.0-py3.8_cuda102_cudnn7_0
torch-geometric-2.1.0.post1-pypi_0
numpy-1.21.5-py38h7a0a035_3
pandas-1.4.4-py38hd77b12b_0
urllib3-1.26.11-pyhd8ed1ab_0
scikit-learn-1.1.1-py38h9235413_0
outdated-0.2.1-pyhd8ed1ab_0

.

> conda info

     active environment : base
            shell level : 1
          conda version : 22.9.0
    conda-build version : 3.22.0
         python version : 3.8.13.final.0
       virtual packages : __cuda=11.7=0
                          __win=0=0
                          __archspec=1=x86_64
  conda av metadata url : None
           channel URLs : https://repo.anaconda.com/pkgs/main/win-64
                          https://repo.anaconda.com/pkgs/main/noarch
                          https://repo.anaconda.com/pkgs/r/win-64
                          https://repo.anaconda.com/pkgs/r/noarch
                          https://repo.anaconda.com/pkgs/msys2/win-64
                          https://repo.anaconda.com/pkgs/msys2/noarch
               platform : win-64
             user-agent : conda/22.9.0 requests/2.28.1 CPython/3.8.13 Windows/10 Windows/10.0.19044
          administrator : False
             netrc file : None
           offline mode : False

Device 0: "NVIDIA GeForce GTX 1070"
  CUDA Driver Version / Runtime Version          11.7 / 10.2
  CUDA Capability Major/Minor version number:    6.1
CUDA Driver = CUDART, CUDA Driver Version = 11.7, CUDA Runtime Version = 10.2, NumDevs = 1 

[rusty1s]

This issue does not seem to be related to pyg-lib. Anyway:

If you do not have PyTorch installed, you need to specify its version as well, e.g.:

conda install pyg pytorch==1.12.0 cudatoolkit=11.3 -c pytorch -c pyg

[moschles]

Quick question. You have double equals on pytorch==1.12.0 but a single equals on cudatoolkit=11.3 Was that intentional?

[moschles]

I have marked this as answer. I proceeded with some other steps.

1. Complete uninstall and removal of extant pytorch.
2. Uninstall of NVIDIA CUDA Toolkit
3. Removal of all extant versions of any kind of Visual Studio

Following those,

1. Install of Visual Studio 2017 Community Edition.
2. Install of NVIDIA CUDA Toolkit 11.3 , such that there are no warnings about Visual Studio of any kind.
3. conda install pyg pytorch==1.12.0 cudatoolkit=11.3 -c pytorch -c pyg

import torch_geometric succeeds and gives version version= 2.1.0 . No other pytorch-geometric or torch_geometric libraries were required, even though there are several floating around, (e.g. Esri 1.7.2).