which version of tensorrt on DrivePX2?Does it support for ONNX?

my PX2 is AutoChauffeur(P2379),
DriveWorks SDK v0.6.67,
Release build with GNU 4.9.2 from v0.6.0-rc7-0-g51bd3aa against Vibrante PDK v5.0.5.0

I find the libnvinfer.so.4.0.2,but slao find the tensorrt pkg 3.0.2-1+cuda9.0 in PX2. So I’m not sure the version of my tensorrt,and weather it support onnx model parsers.

By the way,what’s the different of the following two GPU, I think every PX2 has four GPUs. But how which is suitable for DNN?

GPU devices detected: 2
Device: 0, Graphics Device:( 9) Multiprocessors, (128) CUDA Cores/MP: 1152 CUDA Cores
Device: 1, NVIDIA Tegra X2:( 2) Multiprocessors, (128) CUDA Cores/MP: 256 CUDA Cores

Dear morgan117cai,
Drive 5.0.5 has tensorrt 3.0 which does not support onnx. Onnx support is implemented in tensorrt 4.0.1(https://docs.nvidia.com/deeplearning/sdk/tensorrt-archived/tensorrt_401/tensorrt-release-notes/rel_4-0-1.html#rel_4-0-1). Could you upgrade to latest pdk

Thanks for your rapid reply, Then how can I update my tensorrt on PX2? I think the Drive5.0.5.0b is the latest version, or could I install the tensorrt pkg for PX2 only?

Dear morgan117cai,
Please login to Devzone and download latest SDKManager (https://developer.nvidia.com/nvidia-drive-downloads) to get latest PDK on DrivePX2.