Title：AI and Accelerators: Boom or Bubble?
Date：2024/12/20 14:20-15:30
Location：CSIE R103
Speakers: Prof. Hung-Wei Tseng, University of California, Riverside
Host: Prof.Shang-En Huang
Abstract:
Generative AI (Artificial Intelligence) has reshaped both computer software and hardware. On the software side, service providers invested in developing larger and more effective machine learning models aiming at providing better user experiences. On the hardware side, researchers, large companies and tons of startups innovate on hardware accelerators to support the incredibly fast growing software demand.
However, if we keep following the current trend in focusing on developing new service features, generative AI or in general, AI, is likely to replay the story of the Internet Bubble story. None of the current generative AI service providings demonstrate a profitable business model that is self-sustainable from the perspective of operational costs, carbon dioxide (CO2) emissions, and capital expenditure.
In this talk, Hung-Wei will share research experiences that make the dimming future of generative AI bright. Hung-Wei identified the key to successful and self-sustainable generative AI services must (1) revisit the algorithms in the complete AI/ML pipeline to fully exploit the use of underlying hardware, (2) reuse existing architectural components before designing a new accelerator, and (3) maximize the potential of each hardware accelerator through opportunities that emerging programming models reveal.
Bio:
Hung-Wei is currently an associate professor in the Department of Electrical and Computer Engineering at the University of California, Riverside. He is now leading the Extreme Storage & Computer Architecture Laboratory and focusing on accelerating applications through generalized computing on tensor processors, AI/ML accelerators as well as intelligent data storage systems. He is recognized by facebook faculty research award and IEEE Micro "Top Picks from Computer Architecture" in 2024 and 2020 for his research in designing efficient AI/ML systems. He was previously a visiting researcher at Google’s Tensorflow Lite team and Intel. He got his PhD from the Department of Computer Science and Engineering at the University of California, San Diego.