NVIDIA Jetson DISCUSSION

Upgrading JetPack: which dependencies should be checked before rebuilding an inference application?

Started by sa2016 NVIDIA JetsonJetPack migrationTensorRT compatibilityCUDA dependenciesdeployment testing
3 replies 248 views 4 participants
Latest activity · 27 Sep 2026

Upgrading JetPack: which dependencies should be checked before rebuilding an inference application?

sa2016 NVIDIA Jetson Forum
#1

Community replies 3

Re: Upgrading JetPack: which dependencies should be checked before rebuilding an inference application?

#2

Record the exact Jetson module, carrier, Jetson Linux release and installed runtime versions. Check the target release's support for that hardware and its associated components. The newest JetPack entry is not enough to establish compatibility with every module or custom carrier.

Re: Upgrading JetPack: which dependencies should be checked before rebuilding an inference application?

#3

Keep model sources, engine-build settings, plugins and application dependencies with the release. A serialized TensorRT engine is not a universal model file. Review the compatibility rules for the actual runtime and plan to rebuild artifacts where required.

Re: Upgrading JetPack: which dependencies should be checked before rebuilding an inference application?

#4

Evaluate the upgrade on a separate recoverable image or development unit. Test camera capture, inference accuracy, sustained performance and restart behavior before replacing the working installation. Include custom kernel modules and device-tree changes in the migration record.

TEP COMMUNITY