No visible GPU is found for XGBoost

Hello everyone,

I keep failing at using xgboost with gpu on widows and geforce 1060.
I tried multiple installs, including the rapidsai source.
My GPU and cuda 11.2 work well with tensorflow-gpu, so I guess my setup should be compatible.

nvcc --version
nvcc: NVIDIA ® Cuda compiler driver
Copyright © 2005-2020 NVIDIA Corporation
Built on Mon_Nov_30_19:15:10_Pacific_Standard_Time_2020
Cuda compilation tools, release 11.2, V11.2.67
Build cuda_11.2.r11.2/compiler.29373293_0

My latest env is basically created from scratch:

conda create -n xgboost_env -c nvidia -c conda-forge ipykernel cudatoolkit=11.2 jupyter nb_conda scikit-learn scipy pandas scikit-learn seaborn IPython
pip3 install xgboost

However I keep getting the error:

C:/Users/Administrator/workspace/xgboost-win64_release_1.3.0/src/gbm/ Check failed: common::AllVisibleGPUs() >= 1 (0 vs. 1) : No visible GPU is found for XGBoost.

During the other failures I got the xgboost not compiled with gpu, so I guess my install includes the compiled with gpu support version.

Any advice would be welcome.
Best regards,

Hum, after several hours working on this problem, I made a full reinstall from scratch of cuda and all, and for some reason it works (on my old env that I kept).