top of page
From Application using Toolchain to Chipset’s NPU end to end
Application: AI algorithm, Ex: Image capture + AI detection + BBOX drawing
Tensorflow: TensorFlow post-training quantization
BLAI NN TOOLCHAIN: AI quantization model converted into command stream
Model Interpreter compiles the model and obtain the CPU/NPU computing schedule from BLAI NN TOOLCHAIN. CPU and NPU co-process the AI models. The CPU uses the NN Library operator to infer the model, and the NPU is NN operator accelerator which is much faster than CPU.
bottom of page