AndesAIRE™ AnDLA™ I350/I370 is a deep learning accelerator (DLA) designed to enable high performance-efficient and cost-sensitive AI solutions for edge and end-point inference.
 The very high frequency 1GHz at 28nm
 The very high power-efficiency 5 TOPS/W at 28nm
 Versatile hardware configurable and scalable options (32 to 4096 MAC per core)
 Widely support deep learning frameworks, such as TensorFlow Lite, PyTorch, and ONNX
 Widely support AI CNN/RNN and the cutting-edge Transformer models

Unique design and technologies
 Internal 4-dimensional DMA, local memory, and crossbar bus matrix -> best PPA
 Operator fusion -> best PPA
 General purpose Operator-Set-Architecture (OSA) -> wide model support
 Elegant and scalable architecture design