RK3576, 4 x Cortex-A72 + 4 x Cortex-A53 2.3GHz, 8nm
GPU ARM Mali-G52 MC3
OpenGL ES3.2/OpenCL 2.0/Vulkan1.1
3.0TOPS NPU, INT4,INT8,INT16, float16, Bfloat 16 and tf32 operation
Supports deep learning framework：TensorFlow, Caffe, Tflite, Pytorch, Onnx NN，Android NN, etc.
