Hardware and Software Support for Managed Languages in Data Centers

Explore the future of data center computing in this 59-minute colloquium talk by Martin Maas from UC Berkeley. Delve into the challenges and opportunities presented by managed languages like Java, Python, and Scala in cloud environments. Discover how the shift to serverless deployment models is reshaping the data center stack, enabling custom hardware accelerators, FPGAs, and resource disaggregation. Learn about the concept of a "Holistic" Language Runtime System for better supporting distributed managed-language applications. Examine the potential for custom data center SoCs to provide hardware support for garbage collection. Gain insights into how open-source hardware and the RISC-V ISA are influencing academic computer architecture research, potentially increasing realism and industry adoption. Understand the implications of these advancements for large-scale machine learning, big data analytics, and genomics processing in cloud data centers.

Allen School Colloquia: Martin Maas (UC Berkeley)