The Institute for Energy Performance at UC Santa Barbara presents our 2020 Emerging Technologies Review. In this special year, IEE ETR occurred in a collection of concentrated online Friday workshops in October.
To view the discussions and also panel sessions from these virtual occasions, please click the web links below.
ETR Day 1: Energy Efficient Clouds and Data Centers Workshop
Raj Yavatkar: The Revolution in Data Center Communications and Interconnects
Nicola Peill-Moelter: Hunting Zombies for Sustainability, enjoyable, as well as revenue
Rich Wolski: Energy Efficiency on the (IoT) Edge
Katharine Schmidtke: Cloud-Scale Interconnect Architectures in the Context of Hardware as well as Software Codesign
Daniel Blumenthal: Low Energy Frequency Stabilized Coherent WDM for Data Center Interconnects
Robert Blum: The Promise of Co-Packaged Optics: Paving the Way for Improved Power Efficiency, Size, and also Cost
John Bowers: Photonic Assimilation for Information Centers
William Wang: In The Direction Of Substantially Decreasing the Carbon Footprint of Artificial Intelligence
Panel Session 1: Information Facility as well as Cloud Effectiveness
Mediator: Tim Sherwood
Panelists: Jonathan Balkind, Brent Gorda, Michael Haney, Nicola Peill-Moelter, and also Raj Yavatkar
Panel Session 2: The Optical Change in the Data Center as well as the Cloud
Mediator: John Bowers
Panelists: Alexis Bjorlin, Robert Blum, Gordon Keeler, Kaushik Patel, and also Clint Schow
ETR Day 2: Smart Societal Infrastructure Workshop
Sila Kiliccote: From Demand Reaction to Fleet Electrification: A Browse for High Effect Solutions
Raphaële Clément: Towards the Next Generation of Energy Storage Solutions for the Grid as well as Electric Transportation
Igor Mezic: Improving Building Energy Efficiency through Machine Learning
Gary Barsley: Southern California Edison Grid Updates/ EV Plans
Mahnoosh Alizadeh: Real-Time Control Mechanisms for Community Energy Management
David Erne: Towards a Zero-Carbon Electric Grid
Sangwon Suh: Moving Bits Not Watts
Panel Session: Towards a Smart Societal Infrastructure Built on a Smart Grid
Moderator: Igor Mezic
Panelists: Mahnoosh Alizadeh, Sila Kiliccote, Gary Barsley, as well as David Erne
ETR Day 3: Food-Energy-Water Nexus Workshop
Olivier Jerphagnon - Artificial Intelligence to Solve Practical Problems in the Food as well as Agriculture Sector
Chandra Krintz - The SmartFarm Project
Deborah Perrone - Groundwater Depletion Amplifies the Water-Energy Nexus
Steve DenBaars - Development of Ultraviolet (UV) LED Technology for Disinfection and Sterilization for COVID-19
Stuart Woolf - A Perspective from the California Agricultural Industry
Panel Session: The Food-Energy-Water Nexus
Mediator: Bob Wilkinson
Panelists: Olivier Jerphagnon, Chandra Krintz, Deb Perrone, Steven DenBaars, as well as Stuart Woolf
Wednesday, February 3, 2021
UC Santa Barbara's Institute for Energy Efficiency (IEE) and Facebook are announcing a brand-new collaboration that will accelerate study right into energy-efficient data centers and artificial knowledge (AI). Facebook, a leader in creating, structure, and also running extremely trustworthy as well as effective information centers, will certainly supply a three-year, $1.5 million grant on behalf of the institute's introducing research.
Via this partnership, IEE will examine advanced energy-efficient information facility facilities including low-power optical interconnects for compute networks and also artificial intelligence (ML) with decreased carbon footprint. Devoted to 100% sustainable energy given that 2018, Facebook has actually long believed that the very first step is constantly to utilize the least amount of power feasible by constructing as well as running the most power effective data facilities in the market. Facebook has committed to getting to web zero discharges throughout its value chain in 2030.
" We are deeply grateful to Facebook for their generosity and also assistance of the college as well as the Institute for Energy Efficiency," said Rod Alferness, dean of the College of Engineering. "This present will certainly drive collective discoveries of possibly world-changing options to substantially lower the power needed to drive vital next-generation data facilities as well as applications of artificial intelligence."
" For more than 10 years, Facebook has been concentrated on making, structure and running some of the most effective data facility facilities on the planet. We are thrilled that our study group will partner with UCSB's Institute for Energy Efficiency to aid drive advancement," stated Rachel Peterson, VP of Data Center Strategy at Facebook. "We anticipate interacting to bring power performance to the next degree."
Facebook will certainly likewise assist develop study projects and also offer IEE scientists with insight from their prior experiences creating and running data. The technology leader currently has 8 operational information center websites in the US, with 5 more websites that have been introduced but not yet started operating.
" Facebook is a world leader in information center effectiveness, and we more than happy to partner with them to additional deliver the future generation of technologies to provide efficiency gains," stated IEE Director John Bowers, a distinguished teacher of materials as well as electrical as well as computer system engineering. "They will certainly provide us with the troubles that need to be attended to, and we will collaborate to fix them."
" We are excited to introduce this research study collaboration with IEE," stated Katharine Schmidtke, Facebook's supervisor of sourcing and study lead for the UCSB IEE partnership. "We wish our contribution will certainly aid promote study in data facility power effectiveness to change technology in computer and also communications as well as to drive efficiency in essential work such as AI/ML training and reasoning."
Improving Efficiencies for Data Centers, Artificial Intelligence and Machine Learning
With the partnership powering its study endeavors, IEE anticipates to make substantial headway on 2 of its grand challenges: accomplishing numerous orders-of-magnitude improvement in the effectiveness of both information center as well as AI/ML work.
In the last few years data facilities have actually made significant strides in power performance, but there is still an urgent demand for new innovations to balance out the ongoing growth in need driven by brand-new workloads such as ML, AI, and video clip. Historically, Moore's Law appropriately predicted the ability to counter boosting computing demands with breakthroughs in semiconductor procedure technology, enabling information facilities to scale in a convenient ecological footprint. In current years procedure technology developments have actually reduced as basic physical limits are approached. To address this difficulty, IEE scientists are pursuing a multi-faceted technique, targeting effectiveness renovations in numerous areas. Computer science professor Tim Sherwood as well as various other professors are developing new hardware and design to not just boost the integral performance of processes as well as servers , however additionally lower power made use of for air conditioning. Improvement in algorithmic efficiency is likewise expected to produce substantial decreases in the information facility power footprint connected with AI and ML applications, and this work is being gone after by researchers such as William Wang. The institute's globe course photonics professors, that includes Bowers, are concentrated on opening the following wave of performance innovations connected with interconnects; the essential technologies which provide high bandwidth connections between the compute nodes as well as to users.
" IEE is a globe leader in silicon photonics, which is making it possible for greater capability electronics via co-packaging of electronic devices and also photonics . This research is important to establish the following generation of high-capacity switching chips and becomes a lot more crucial for future generations of electronics," said Bowers, an internationally prominent authority on silicon photonics and also optoelectronics.
" A present of this scale is very rare and also absolutely transformative ," claimed Wang, the Mellichamp Chair of Artificial Intelligence and also Designs. "Last summertime, helpful site we worked together with the Facebook AI Group to build an reasoning engine for Natural Language Processing that attained 10 times much more energy performance than the present design. It further shows the value as well as potential influence of collaborating with industrial partners to recognize real-world issues as well as connect clinical research study and design."
In acknowledgment of Facebook's gift, the university will certainly name one of the speculative information center laboratories in Henley Hall the "Facebook Data Center Energy Efficiency Lab." A devotion as well as recognition event will certainly be scheduled at a later day, as soon as state as well as county public wellness limitations related to the coronavirus are raised.
We are thrilled that our research group will certainly partner with UCSB's Institute for Energy Efficiency to help drive innovation," claimed Rachel Peterson, VP of Data Center Strategy at Facebook. "We look ahead to functioning with each other to bring energy performance to the following level."
In current years data facilities have actually made significant strides in power effectiveness, but there is still an immediate need for brand-new innovations to offset the ongoing growth in need driven by brand-new workloads such as ML, AI, and also video. Improvement in mathematical efficiency is additionally expected to produce considerable reductions in the information center power impact connected with AI and also ML applications, as well as this work is being gone after by researchers such as William Wang. "Last summer season, we worked together with the Facebook AI Group to construct an reasoning engine for Natural Language Processing that achieved 10 times much more power effectiveness than the existing design.