Jetson Orin Nano shows the NVIDIA-Orin architecture with NVIDIA ampere architecture GPU.

It has up to 8 streaming multiprocessors (SM) and consists of 1024 CUDA cores and up to 32 tensor cores for AI processing.

The third-generation Tensor Cores of NVIDIA Ampere Architecture outperforms the previous generation in performance per watt and brings higher performance while supporting sparsity.

With sparsity, you can double the throughput of Tensor Core operations by taking advantage of the fine-grained structured sparsity in deep learning networks.

To speed up all parts of the application pipeline, Jetson Orin Nano also includes a 6-core Arm Cortex-A78AE CPU, video decoding engine, ISP, video image synthesizer, audio processing engine, and video input block.