CUDA driver version is insufficient for CUDA runtime (

Error:

Check failed: error == cudaSuccess (35 vs. 0) CUDA driver version is insufficient for CUDA runtime version

I have tried to follow different solutions in this forum with similar error as it’s common but no success. I am using latest drivers / toolkit / cudnn on Ubuntu 20.04.1

nvidia-smi:

nvcc:
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Tue_Mar__8_18:18:20_PST_2022
Cuda compilation tools, release 11.6, V11.6.124
Build cuda_11.6.r11.6/compiler.31057947_0

Just to give more bits of information, I have compiled a C++ code but I can’t run (execute) the demo example as I get this insufficiency error. However, the same code compiles and executes using google colab which uses CUDA 11.2

Working colab code for reference to check boilerplate: Google Colab

Worth to note that I can use frameworks that support CUDA, say using PyTorch and have a Tensor in a CUDA device. No issues. Maybe that’s due to PyTorch’s CUDA toolkit?