Using CUDA 12.2 on Xavier Dev Board with Jetpack 5.1.2

Robotics & Edge Computing Jetson & Embedded Systems Jetson AGX Xavier
1

Hello,

When we explore the CUDA UMD and CUDA Toolkit version compatibility table for NVIDIA JetPack 5.x we observe that JetPack 5.1.2 is compatible with CUDA 12.2.
We can update CUDA 11.4 on JetPack 5.1.2 to CUDA 12.2 using the download scripts.

After that, we can build Torch from source (without TensorRT) and make training and inference.
Moreover, we can make inference using onnxruntime.

But we cannot find a way to build tensorrt compatible with CUDA 12.2 on Xavier Dev board.
We could not find compatible Docker or TensorRT GA.
Is this not supported?

Can we use CUDA 12.2 on Xavier Dev Board even if we do not use TensorRT for inference?

Or we do not any update other than CUDA 12.2 update since it is written in Jetpack 5.1.2 documentation that Starting with JetPack 5.0.2, upgrade to latest and greatest CUDA releases from CUDA 11.8 onwards without the need to update Jetson Linux other JetPack components.??

3

Hi,

Looks like this duplicates to this topic.
Let’s discuss the following there and close this one.

Thanks.