Error reported in import isaacGym: RuntimeError: Found no NVIDIA driver on your system

Hello,

My isaacgym seems to be unable to find the correct graphics card driver. I obtained different results by executing the following two commands in the same conda environment:

# torch: 2.2.1+cu121

python -c "import isaacgym; import torch; print(torch.zeros(1).cuda()); print(torch.cuda.is_available()); torch._C._cuda_init()"

Importing module 'gym_38' (/home/ubuntu/1Package/isaacgym/python/isaacgym/_bindings/linux-x86_64/gym_38.so)
Setting GYM_USD_PLUG_INFO_PATH to /home/ubuntu/1Package/isaacgym/python/isaacgym/_bindings/linux-x86_64/usd/plugInfo.json
Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/home/ubuntu/.conda/envs/ikloco/lib/python3.8/site-packages/torch/cuda/__init__.py", line 302, in _lazy_init
    torch._C._cuda_init()
RuntimeError: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx

python -c "import torch; print(torch.zeros(1).cuda()); print(torch.cuda.is_available()); torch._C._cuda_init()"

tensor([0.], device='cuda:0')
True

I also made some other attempts

# torch: 2.1.1+cu118 or 2.2.1+cu118 or 2.2.0+cu118 or 1.13.1+cu117
# 

python -c "import isaacgym; import torch; print(torch.zeros(1).cuda()); print(torch.cuda.is_available()); torch._C._cuda_init()"

Importing module 'gym_38' (/home/ubuntu/1Package/isaacgym/python/isaacgym/_bindings/linux-x86_64/gym_38.so)
Setting GYM_USD_PLUG_INFO_PATH to /home/ubuntu/1Package/isaacgym/python/isaacgym/_bindings/linux-x86_64/usd/plugInfo.json
Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/home/ubuntu/.conda/envs/ikloco/lib/python3.8/site-packages/torch/cuda/__init__.py", line 302, in _lazy_init
    torch._C._cuda_init()
RuntimeError: Unexpected error from cudaGetDeviceCount(). Did you run some cuda functions before calling NumCudaDevices() that might have already set an error? Error 34: CUDA driver is a stub library

python -c "import torch; print(torch.zeros(1).cuda()); print(torch.cuda.is_available()); torch._C._cuda_init()"

tensor([0.], device='cuda:0')
True

When I tried 1.10.0+cu113 (which ran well on my other three machines), both lines of command returned the same error

# torch: 1.10.0+cu113

Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/home/ubuntu/.conda/envs/ikloco/lib/python3.8/site-packages/torch/cuda/__init__.py", line 214, in _lazy_init
    torch._C._cuda_init()
RuntimeError: Unexpected error from cudaGetDeviceCount(). Did you run some cuda functions before calling NumCudaDevices() that might have already set an error? Error 34: CUDA driver is a stub library

Could you kindly give me some advise about where should I troubleshoot this problem?

My other info:
Ubuntu 20.04
GeForce RTX 4080
Driver Version: 535.129.03 (I have also tried 555)
Isaacgym Preview 4