If you have an event you would like posted, please contact email@example.com
- This event has passed.
Seminar: Reconfigurable Computing for Artificial Intelligence
August 8, 2018 @ 10:30 am - 11:30 am
Speaker: Prof. Shao-Jun Wei, Director of Institute of Microelectronics, Tsinghua University
Title: Reconfigurable Computing for Artificial Intelligence
Time: 10:30 – 11:30am on Wednesday, August 8, 2018
Location: LAS 1002
The age of artificial intelligence (AI) is coming. However, there is still a long way to go from intelligence augmentation (IA) to AI. As we all know, there are three important factors for artificial intelligence to rise again, namely algorithm, data and computing power. It is no doubt that AI relies on chip technology. But for engineers engaged in AI chip development, they still face many challenges. At first, artificial intelligence algorithms are still evolving rapidly. Secondly, one algorithm corresponds to an application, and there is no unified algorithm yet. In this case, it is not appropriate to use traditional ASIC architecture to implement AI chips. In addition, the AI chip, especially the AI chip used to terminals, needs high energy efficiency. An AI chip should not only be able to achieve a computing power of trillion operations per second (TOPS), but also achieve TOPS/W energy efficiency. Obviously, finding a new architectures is the most important way for AI chips design. Reconfigurable computing technology, also known as software defined chip technology, is a new architecture. It has not only high flexibility like CPU, but also high performance and high energy efficiency like ASIC, and most importantly, its architecture and function change in real time with the change of software (application). It is considered as an ideal architecture for AI chip.
Tsinghua University has been working in the research of reconfigurable computing technology for more than 10 years. It has published 150+ papers in this field, with 100+ Chinese and US patents, and achieved good results. This report attempts to analyze the challenges faced by AI chip. By introducing the reconfigurable computing technology, we try to point out the methods and ways to realize the AI chip using this architecture so that audiences can have a deeper understanding of the reconfigurable computing.
Prof. Shao-Jun Wei is the Director of Institute of Microelectronics, Tsinghua University; Chief Scientist of the State Key Science and Technology Project; Member of the National Integrated Circuit Industry Development Advisory Committee; President of Fabless Chapter, and Vice President of China Semiconductor Industry Association (CSIA); Prof. Wei was the President & CEO of Datang Telecom Technology Co., Ltd. and the CTO of Datang Telecom Industry Group.
Prof. Wei has been working on VLSI design methodologies research and reconfigurable computing technology research. He has published more than 200 papers and 3 monographs. He is the Fellow of Chinese Institute of Electronics (CIE) and Senior Member of IEEE.
Prof. Wei had won many awards including National Second Award for Technology Invention, National Second Award for Technology Progress, SIPO & WIPO Patent Golden Award, and First Award for Science and Technology of Ministry of Education, as well as First Award for Technology Invention of CIE, and etc.