Cuda

Accelerated Computing GPU-Accelerated Libraries
1

Hello NVIDIA team,

I am working on a real-time computer vision project on a Jetson device running JetPack 6.x (L4T r36.x), and I need to use CUDA-accelerated PyTorch, not the CPU-only version.

Before proceeding further, I would like to confirm the officially supported installation method directly from NVIDIA.

>> What is the officially supported method from NVIDIA to install GPU-enabled PyTorch on JetPack 6.x?

I specifically need PyTorch with:

  • CUDA support

  • cuDNN support

  • Full GPU inference (torch.cuda.is_available() == True)

I want to ensure that I am following the correct, NVIDIA-endorsed approach rather than relying on unofficial or incomplete installation

2

I suggest asking questions like this on the relevant Jetson forum. (such as this)

You will probably be directed to use a NGC container.