Edge and IoT-supported Intelligent Augmented Reality: Promise, Challenges, and Solutions
Title: Edge and IoT-supported Intelligent Augmented Reality: Promise, Challenges, and Solutions
Speaker: Maria Gorlatova
Abstract: Mobile augmented reality (AR), which integrates virtual objects with 3D real environments in real time, has been showing outstanding potential in many application domains including education, retail, and healthcare. AR is broadly expected to redefine how we interact with technology and the world around us. Yet current AR falls short of many of the expectations. This talk presents our vision of multi-device, edge computing-supported and Internet-of-Things (IoT)-integrated platforms for next-generation intelligent context-aware AR. I describe shortcomings in modern AR platforms’ spatial and semantic awareness capabilities and identify key gaps that need to be addressed to enable AR to become robust and resource-efficient; I present solutions to some of the challenges, touching on topics in edge computing, machine learning, and resource-efficient simultaneous localization and mapping (SLAM). I also highlight opportunities associated with close integration of AR platforms and their users and describe a set of solutions for robust gaze-based user context awareness in AR. The talk showcases several applications of next-generation context-aware AR, such as AR-based surgical guidance and AR support for mental and behavioral health and well-being.
Short Bio: Dr. Maria Gorlatova is a Nortel Networks Assistant Professor of Electrical and Computer Engineering and Computer Science at Duke University. Her research is in emerging pervasive technologies, with a focus on next-generation augmented reality and the Internet of Things. She received her Ph.D. in Electrical Engineering from Columbia University, and her M.Sc. and B.Sc. (Summa Cum Laude) degrees in Electrical Engineering from University of Ottawa, Canada. She spent two years at Princeton University Electrical Engineering Department as an Associate Research Scholar and an Associate Director of the Princeton University EDGE Lab. Dr. Gorlatova was named among the 10 N2Women Rising Stars in 2019, and has received multiple awards including the 2021 NSF CAREER Award, 2021 Meta Research Award, 2023 CISCO Research Award, ACM SenSys Best Student Demonstration Award, ACM/IEEE IPSN Best Research Artifact Award, Google Anita Borg Fellowship, the IEEE Communications Society Young Author Best Paper Award, and the IEEE Communications Society Award for Advances in Communications.
Towards Performant Virtualization of Floating Point Arithmetic
Title: Towards Performant Virtualization of Floating Point Arithmetic
Speaker: Peter Dinda
Abstract: Using alternative arithmetic systems within an existing scientific codebase that is written to (and compiled for) the IEEE standard is a major challenge. The NSF CSR-supported Buoyancy Project is exploring how to address this challenge through virtualizing floating point hardware, starting on x64. The goal of the floating point virtual machine (FPVM) is to allow an existing application binary to be seamlessly extended to support the desired alternative arithmetic system with overheads determined by that system and not the virtualization mechanisms. I will describe the current FPVM trap-and-emulate-based implementation, including newly-developed kernel support that reduces one of the key overheads by 2-3x. There remains much to be done, however. If time allows, I will also touch on other aspects of the Buoyancy Project, such as evaluating user understanding of IEEE floating point, efficient monitoring of applications, and very early work on attacks specific to scientific applications.
Short Bio: Peter Dinda is a professor in the Department of Computer Science at Northwestern University, and also holds an appointment in the Department of Electrical and Computer Engineering. He headed the Computer Engineering and Systems division for five years within the previous Department of Electrical Engineering and Computer Science. He holds a B.S. in electrical and computer engineering from the University of Wisconsin and a Ph.D. in computer science from Carnegie Mellon University. He works in experimental computer systems, particularly parallel and distributed systems, and has authored over 140 scientific papers, authored or is a major contributor to several large publicly available codebases, and holds five patents. His research currently involves virtualization and operating systems for distributed and parallel computing, programming languages for parallel computing, resilience of floating point arithmetic, and individualized privacy in IoT systems. He is a Fellow of the IEEE. You can find out more about him and the Buoyancy Project at pdinda.org.
Generalized Caching as a Service
Title: Generalized Caching as a Service
Speaker: Raju Rangaswami
Abstract: Data centers today host large numbers of workloads and many of these workloads consume significant storage resources. Given the long history of successes in storage caching, it is only natural such successes bear fruit in modern data centers, at scale. In this talk, we motivate a new approach for building a generalized caching service for cloud data centers. Departing from existing application, storage, or data-type specific caches, this service unifies and abstracts data center caching resources making these available to any workload and for any data type. Also departing from past caching practices, this caching service is fault-tolerant, allowing it to cache writes without risk of data loss. Finally, as expected from production storage systems, this caching service also implements per-workload performance guarantees.
Short Bio: Dr. Raju Rangaswami is a Professor of Computer Science at Florida International University where he directs the Systems Research Laboratory. His work has focused on computer systems and software, including operating systems, distributed systems, storage systems, computer security, and real-time systems as well as application domains such as web services, databases, cloud computing, and mobile computing. He is a recipient of the NSF CAREER award and the Department of Energy CAREER award. His research is also supported by industry entities including IBM, Intel, NetApp, and Seagate.
Offloading Intra-Server Orchestration to Smart NICs
Title: Offloading Intra-Server Orchestration to Smart NICs
Speaker: Aditya Akella
Abstract: Orchestrating requests at a datacenter server entails load balancing and scheduling requests belonging to different services across CPUs, and adapting CPU allocation to request bursts. It plays a central role in meeting tight tail latency requirements and ensuring high throughput and optimal CPU utilization. Today’s server-based orchestration approaches are neither scalable nor flexible. In this talk, I will argue for offloading orchestration entirely to the server’s network interface card (NIC). I will present RingLeader, a new programmable “smart” NIC with novel hardware units for software-informed request load balancing and programmable scheduling, and a new light-weight OS-NIC interface that enables close NIC-CPU coordination and supports NIC-assisted CPU scheduling. I will conclude my talk with examples of other ways that smart NICs are changing the landscape of data center computing.
Short Bio: Aditya Akella is a Regents Chair Professor of Computer Science at UT Austin. Aditya received his B. Tech. from IIT Madras (2000), and Ph.D. from CMU (2005). His research spans computer systems and networking, focusing on programmable networks, formal methods in systems, and systems for big data and machine learning. His work has influenced the infrastructure of some of the world’s largest online service providers. Aditya has received many awards for his contributions, including the ACM SIGCOMM Test of Time Award (2022), selection as a finalist for the US Blavatnik National Award for Young Scientists (2020 and 2021), UW-Madison “Professor of the Year” award (2019 and 2017), IRTF Applied Networking Research Prize (2015), SIGCOMM Rising Star award (2014), NSF CAREER award (2008), and several best paper awards.
Decarbonizing Cloud Computing Using CarbonFirst
Title: Decarbonizing Cloud Computing Using CarbonFirst
Speaker: Prashant Shenoy
Abstract: In this talk, I will discuss our CarbonFirst approach to designing sustainable cloud computing systems. Our goal is to make carbon efficiency a first-class design metric, similar to traditional metrics of performance and reliability. I will explain how today’s systems can be made first carbon-aware by exposing energy and carbon usage information to software platforms and then made carbon-efficient by providing control over the system’s carbon usage. I will present application case studies on how modern cloud applications can employ these mechanisms to reduce their carbon footprint. I will end with some open research questions in the field of sustainable computing.
Short Bio: Prashant Shenoy is currently a Distinguished Professor in the College of Information and Computer Sciences at the University of Massachusetts Amherst. He received the B.Tech degree in Computer Science and Engineering from the Indian Institute of Technology, Bombay and the M.S and Ph.D degrees in Computer Science from the University of Texas, Austin. His research interests lie in distributed systems and networking, with a recent emphasis on cloud and sustainable computing. He has been the recipient of several best paper awards at leading conferences, including a Sigmetrics Test of Time Award. He is a fellow of the ACM, the IEEE, and the AAAS.
Attention-Driven Software Architecture for Autonomous Robotic Agents
Title: Attention-Driven Software Architecture for Autonomous Robotic Agents
Speaker: Hyoseung Kim
Abstract: While there has been significant progress in developing autonomous systems, challenges still remain in effectively processing vast amounts of sensor data and making timely decisions, especially for small robotic agents with limited computing power. In this talk, I will introduce our recent project aimed at addressing this issue. The primary goal of this project is to create an attention-driven software architecture that can identify and prioritize critical information from sensors, enabling timely decision-making while considering resource constraints and uncertainties in the environment. This architecture is designed to holistically optimize computation scheduling, perception, and planning, with capabilities to adapt to changes in context and anticipate future actions. Our research tasks involve context adaptive scheduling of autonomous computation pipelines, learning-based perception to anticipate future actions in dynamic environments, and motion planning and decision making based on anticipated actions in the presence of uncertainty. I will share preliminary results motivating our research, and discuss expected technical accomplishments and their potential to tackle fundamental challenges associated with time-sensitive scenarios in resource-constrained autonomous systems.
Short Bio: Hyoseung Kim is an Associate Professor in the Department of Electrical and Computer Engineering at the University of California, Riverside (UCR). He is also serving as the Chair of the Computer Engineering Program at UCR. He received the PhD degree in Electrical and Computer Engineering from Carnegie Mellon University in 2016, and the MS and BS degrees in Computer Science from Yonsei University, Korea, in 2007 and 2005, respectively. His research interests are in real-time embedded and cyber-physical systems, autonomous systems, and smart sensing, with a focus on the intersection of systems software, hardware platforms, and analytical techniques. His research projects have been supported by NSF, ONR, DoD, DoJ, USDA/NIFA, etc. He is a recipient of the NSF CAREER Award and the Fulbright Scholarship Award. His research contributions have been recognized with Best Paper Awards at RTAS and RTCSA, and Best Paper Nominations at EMSOFT and ICCPS. For more information, please visit https://www.ece.ucr.edu/~hyoseung/.
Mosaic Pages: Increasing TLB Reach with Reduced Associativity Memory
Title: Mosaic Pages: Increasing TLB Reach with Reduced Associativity Memory
Speaker: Donald E. Porter
Abstract: The TLB is increasingly a bottleneck for big data applications. In most designs, the number of TLB entries are highly constrained by latency requirements and growing much more slowly than the working sets of applications. Many solutions to this problem, such as huge pages, perforated pages, or TLB coalescing, rely on physical contiguity for performance gains, yet the cost of defragmenting memory can easily nullify these gains. This talk introduces mosaic pages, which increase TLB reach by compressing multiple, discrete translations into one TLB entry. Mosaic leverages virtual contiguity for locality but does not use physical contiguity. Mosaic relies on recent advances in hashing theory to constrain memory mappings, in order to realize this physical address compression without reducing memory utilization or increasing swapping. Our results show that Mosaic’s constraints on memory mapping do not harm performance and reduce TLB misses in several workloads by 6-81%.
Short Bio: Don Porter is a Professor of Computer Science at the University of North Carolina at Chapel Hill. Porter’s research interests broadly involve developing more efficient and secure computer systems. Porter earned a Ph.D. and M.S. from The University of Texas at Austin, and a B.A. from Hendrix College. He has received awards including the NSF CAREER Award, the Bert Kay Outstanding Dissertation Award from UT Austin, an ASPLOS Distinguished Paper Award in 2023, an ASPLOS Influential Paper Award in 2022, and Best Paper Awards at FAST 2016, EuroSys 2016, and RTNS 2018. Don Porter is a Professor of Computer Science at the University of North Carolina at Chapel Hill. Porter’s research interests broadly involve developing more efficient and secure computer systems. Porter earned a Ph.D. and M.S. from The University of Texas at Austin, and a B.A. from Hendrix College. He has received awards including the NSF CAREER Award, the Bert Kay Outstanding Dissertation Award from UT Austin, an ASPLOS Distinguished Paper Award in 2023, an ASPLOS Influential Paper Award in 2022, and Best Paper Awards at FAST 2016, EuroSys 2016, and RTNS 2018.
Architecting Computer System Abstraction with Secure Environment in Mind
Title: Architecting Computer System Abstraction with Secure Environment in Mind
Speaker: Yan Solihin
Abstract: In this talk, I will point out that current Trusted Execution Environments (TEE) abstractions of secure enclaves are incompatible with traditional system abstraction of compute (processes and threads) and data (shared memory, files, etc.), making it hard to adopt TEE universally. I will discuss that more research is needed to bring TEE into compatibility with traditional system abstraction and challenges in achieving it.
Short Bio: Yan Solihin is the Director of Cybersecurity and Privacy Cluster, and Charles N. Millican* Professor of Computer Science at University of Central Florida. He obtained his Ph.D. in computer science from the University of Illinois at Urbana-Champaign (UIUC) in 2002. His research interests include computer architecture and system, and secure processors. He is a recipient of 2023 HPCA Test of Time Award, 2010 and 2005 IBM Faculty Partnership Award, 2004 NSF Faculty Early Career Award, and 1997 AT&T Leadership Award. He was one of pioneers in cache sharing fairness and Quality of Service (QoS), efficient counter mode memory encryption, and Bonsai Merkle Tree, which have significantly influenced Intel Cache Allocation Technology and Secure Guard eXtension (SGX). He received IEEE Fellow “for contributions to shared cache hierarchies and secure processors” in 2017. He is listed in the HPCA Hall of Fame, ISCA Hall of Fame, and Computer Architecture Total (CAT) Hall of Fame.
Bringing Foundational Models to Consumer Devices via ML Compilation
Title: Bringing Foundational Models to Consumer Devices via ML Compilation
Speaker: Tianqi Chen
Abstract: Deploying deep learning models on various devices has become an important topic. Machine learning compilation is an emerging field that leverages compiler and automatic search techniques to accelerate AI models. ML compilation brings a unique set of challenges: emerging machine learning models; increasing hardware specialization brings a diverse set of acceleration primitives; growing tension between flexibility and performance. In this talk. I then discuss our experience in bringing foundational models to a variety of devices and hardware environments through machine learning compilation.
Short Bio: Tianqi Chen is currently an Assistant Professor at the Machine Learning Department and Computer Science Department of Carnegie Mellon University. He is also the Chief Technologist of OctoML. He received his PhD. from the Paul G. Allen School of Computer Science & Engineering at the University of Washington. He has created many major learning systems that are widely adopted: XGBoost, TVM, and MLC-LLM.
Systems Research in Quantum Computing
Title: Systems Research in Quantum Computing
Speaker: Frank Mueller
Abstract: Quantum computing has become reality with the deployment of different device technologies accessible through the cloud. However, current hardware technologies pose a number of problems, which require research advances in the systems are of quantum. This talk provides an overview of contemporary problems, sample solutions and open research challenges in this filed. It also highlights benefits of interdisciplinary collaborations and discusses funding opportunities.
Short Bio: Frank Mueller (mueller@cs.ncsu.edu) is a Professor in Computer Science and a member of multiple research centers at North Carolina State University. Previously, he held positions at Lawrence Livermore National Laboratory and Humboldt University Berlin, Germany. He received his Ph.D. from Florida State University in 1994. He has published papers in the areas of quantum computing, parallel and distributed systems, embedded and real-time systems, and compilers. He is a member of ACM SIGPLAN, ACM SIGBED and an ACM Fellow as well as an IEEE Fellow. He is a recipient of an NSF Career Award, an IBM Faculty Award, a Google Research Award and two Fellowships from the Humboldt Foundation.
