Update Tensorrt to 7.1.3 without reflashing new jetpack

Hello,

I have over 50 jetson nano boards that are on Jetpack 4.2 [L4t 32.1.0].
Currently they have
CUDA: 10.0.166
TensorRT: 5.0.6.3
CuDNN: 7.3.1.28

My new deep learning app was built on Jetpack4.4.1 installation and has

TensorRT 7.1.3.0
CuDNN 8.0.0.180
CuDA 10.2.89

Since it is not feasible to flash all 50 boards with new jetpack 4.4.1.

Can you suggest some way to update the TensorRT versions on all jetson nano to 7.1.3 without reflashing, so that new code could work on these boards?

Thanks

Hi,

We do have an OTA feature, but the support starts from JetPack 4.3.
So you will need to reflash it to update the software and library version.

Thanks.

Is it possible to download the deb packages from somewhere and install them?

I saw dusty_nv mention something like this in one of the other threads. Setting up CUDA/TensorRT manually on Jetson AGX Xavier

Hi,

To upgrade TensorRT 5.0 into v7.1, you will need to install the corresponding GPU driver also.
However, Jetson’s GPU driver integrates into the OS, which means you still need to reflash the device.

After JetPack4.3, you can upgrade OS and components with our OTA feature directly.

Thanks.