@eracle94 to test TRT, I would run this with an actual model, i.e. trtexec --onnx=my_model.onnx (or you can specify a caffe model as well) Is the jetson-inference container still not running for you either?
i tried to install the 5.1.1 but it says that i have no space available ( but it boots up with ubuntu/nvdia os and i can use as a normal pc . meanwhile is trying to install cuda drivers and rest of the stuff ) . this is impossible because i have 64gb sd card; maybe the sdkmanager is partitioning my sdcard ? on the other sdcard iwth 5.0.2 i didn’t have this issue .
also i tried to re use the 5.0.2 version ( installed on another sd card ) doesn’t boot up , i guess because the board is trying to boot up from the internal storage . but if i try to press ‘esc’ to enter the boot manager , it doesn’t recognize the input , even if i press " f11 " , it was working " perfectly " before flashing the 5.1.1 in other sd card
i have no idea why it is so difficult to just have a board working and following your tutorial
If it wasn’t for the custom carrier board you are using, which I’m unsure what the flashing procedure is for, I’d say that you don’t need SDK Manager at all and can simply flash the SD card image with Etcher tool. The SD card images come with CUDA/cuDNN/TensorRT already pre-installed. Have you tried that? https://developer.nvidia.com/downloads/embedded/l4t/r35_release_v3.1/sd_card_b49/jp511-xnx-sd-card-image.zip/
Typically when you first boot up a new image and setup your user, near the end it will prompt you to resize the partition on the SD card to it’s maximum space.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.