Abstract
This year, the Institute for Energy Efficiency at UC Santa Barbara held two virtual workshops as part of the 2022 Emerging Technologies Review.
The ETR workshops covered new developments in: Energy Efficiency in Buildings in the Post-COVID-19 Era on January 20th, and Future Data Center Architectures and Increased Energy Efficiency on January 21st. Each virtual workshop featured world class experts and leaders in that arena from industry, academia, and government.
2022 ETR Speakers
Day 1: Energy Efficiency in Buildings in the Post-COVID-19 Era
Agenda
There are large gains in energy efficiency in buildings that have been achieved by a combination of more efficient equipment and AI-driven automation. The COVID-19 pandemic has brought forward the focus on health aspects that the industry is solving in order to provide a safe environment for staff. This workshop focused on the interplay between the two, specifically how the increased needs on the health side and the consequent increase in energy expenditure can be mitigated by further deployment of automation and AI.
Speakers
Mary Ann Piette, LBNL: Energy Efficiency in Buildings and the Urgency for Grid Integration and Decarbonization
Martin Fischer, Stanford University: Performance by Design
Tim Salsbury, PNNL: Post-Pandemic Opportunities for Building Control and Automation for a Clean Energy Future
Dan Nall, ASHRAE Fellow: Building Operations Decarbonization: 2022
Mead Rusert, Automated Logic Corporation: Comfort, Health, and Energy - Finding the New Balance
Day 2: Future Data Center Architectures and Increased Energy Efficiency
Agenda
This virtual workshop will focus on architecture, communications and systems issues driving energy efficiency in the cloud and the data center. It will include talks from industry leaders including Katharine Schmidtke, Bharath Muthiah and Rob Stone from Meta Platforms, Robert Blum from Intel, Robert Ober and Larry Denisson from NVIDIA, Manish Mehta from Broadcom, and talks from UCSB Professors John Bowers, Clint Schow, Yufei Ding, and Lei Li. It will include two panel sessions, one focused on cloud systems and architectural issues and the other on photonic data center communications and interconnects.
Speakers
Welcome Address and Introduction to the Institute for Energy Efficiency
Katharine Schmidtke, Meta: Problems Faced by Data Centers Today and Future Needs
Bharath Muthiah, Meta: AI Use Cases and Deployments Today, and Problem Statement for Future
Robert Ober, NVIDIA: AI Cluster Trends
Cliff Young, Google: Accelerators are Green; Cloud Accelerators are Greener
Frank Helms, AMD: Trends in Accelerated Computing
Yufei Ding, UCSB: Compiler and Runtime Support for Exploring AI-GPU Acceleration
Surendra Anubolu, Broadcom: Energy Efficient Training Fabric
Lei Li, UCSB: High-Performance Training and Inference on GPUs for NLP Models
Round Table Discussion: AI/ML System and Workload Optimization
Rob Stone, Meta: Optical Interconnects in Data Centers: Drivers and Application Spaces
Ram Huggahalli, Microsoft: Co-Packaged Optics in Future Server Designs
Manish Mehta, Broadcom: Scaling Highly Integrated Optical Interconnects
Thomas Liljeberg, Intel: High Volume Silicon Photonics for Co-Packaged Optics and Optical I/O
Chris Cole, Parallax Group: Data Center Energy Savings in this Decade
Clint Schow, UCSB: Bringing Low-Power O-Band Coherent Optics to the Datacenter
Ravi Agarwal, Meta: Meta Data Centers Heterogenous Integration Driven by AI/ML and Network Applications
John Bowers, UCSB: High Capacity, Energy Efficient Interconnects for Data Centers
Round Table Discussion: Interconnect Optimization