Research on Heterogeneous architecture 

Aiming at the problems of low performance and high energy consumption of a large number of emerging artificial intelligence applications and big data analysis applications. The team study the heterogeneous processors customized according to the application characteristics, such as CPU, GPU and FPGA, and the high-performance and low-energy heterogeneous processing platform constituted by these processors. 

Heterogeneous reconfigurable artificial intelligence processor chip design

Research on the design method, architecture, EDA tools, and circuit optimization of reconfigurable processors for the next generation of artificial intelligence technology, and verify the prototype with FPGA and custom ASIC chip implementation. The project team has completed the design and streaming of the three-board intelligent processor, presided over a major special project of "agile customizable intelligent visual processor and system application" in Guangdong Province, and carried out intelligent algorithms for hybrid artificial neural networks, reinforcement learning, and Gaussian processes. Custom architecture design work.