Deep Learning Accelerator (DLA)
This is something invented by NVIDIA.
This exists onboard the Jetson Orin!!
It’s a custom hardware accelerator for deep learning applications.
To use it:
TensorRT does not run on DLA automatically!
TensorRT doesn’t run out of the box on the DLA! You need to configure it to do so.
https://github.com/felixchenfy/Monocular-Visual-Odometry/blob/master/CMakeLists.txt