How to update tensorrt for my nvidia nx to 7.2?

[fanyj233]

i can transfrom my model of onnx to tensorrt using DLA only on my computer, but it can not work on nvidia nx.
i found there was different with tensorrt’s version.my computer is 7.2.2.3 .and nx is 7.1.3.0.
computer can work well,but nx print error next:

[TensorRT] ERROR: Try increasing the workspace size with IBuilderConfig::setMaxWorkspaceSize() if using IBuilder::buildEngineWithConfig, or IBuilder::setMaxWorkspaceSize() if using IBuilder::buildCudaEngine.
[TensorRT] ERROR: ../builder/tacticOptimizer.cpp (1715) - TRTInternal Error in computeCosts: 0

but i want run it on nx instead of my PC.
what can i do ?

[spolisetty]

Hi @fanyj233,

TensorRT engine is not portable. To run on XavierNX platform, you will need to create the engine on the XavierNX directly.
For your reference,

Run out of memory when creating TensorRT engine from onnx model

I have a custom PyTorch model, which I exported to onnx format (1.3 gb) Now I wanted to run it on Jetson Xavier NX using TensorRT. When trying to create an trt engine (similar to the yolov3 example provided on Jetson Xavier Nx in /usr/src/tensorrt/samples/python/yolov3_onnx/onnx_to_tensorrt.py) I’m getting the following errors: [TensorRT] ERROR: Try increasing the workspace size with IBuilderConfig::setMaxWorkspaceSize() if using IBuilder::buildEngineWithConfig, or IBuilder::setMaxWorkspaceSi…

If you any further assistance, we recommend you to post your query on related forum.

Jetson Xavier NX

Thank you.

[fanyj233]

thanks @spolisetty ，
i know it is not portable.so i want to transfrom my models on nx,but it failed.
but it can be trandfromed well in my PC.
there are different version of my PC and nx.
so i need to know how to update my nx’s tensorrt to 7.2.2.3 as my PC.

[spolisetty]

Hi @fanyj233,

Latest version available for NX is 7.1.3 only.

JetPack SDK

Thank you.