Seminars

Monday, September 25, 2023
10:20 a.m. – 11:10 a.m. (CST)
ETB 1034

Laxmikant (Sanjay) Kale
Director & Research Professor at the Parallel Programming Laboratory
Paul and Cynthia Saylor Professor Emeritus of Computer Science
University of Illinois Urbana-Champaign

Title: “The Migratable Objects Parallel Programming Model: Successes and Prospects”

Talking Points: 

• Parallel Programming Model with Runtime Adaptivity
• Automated Dynamic Load balancing and energy optimization
• Highly Scalable Parallel Applications
• Coronavirus simulation on supercomputers

Abstract
The Migratable Objects programming model (MOPM) represents an approach to parallel programming where the notion of a processor is virtualized, and represented by an encapsulated object that can be migrated to any physical processor or host at will by an intelligent runtime system. Combined with over-decomposition, it separates concerns about how to partition data and what computations to do in parallel from where the data resides and which processor executes which actions. Thereby, it empowers highly adaptive runtime systems, which supports asynchronous task-based models and uniquely (and most consequentially) dynamic load balancing. It automatically overlaps communication and computation overlap and engenders parallel composition of independent modules efficiency. MOPM also supports automatic power and energy related optimizations as well as fault tolerance.

I will review the basic ideas of the programming model, its baseline implementation in Charm++, and the successes it has notched. The well-known application NAMD, which was used in many highly scaled supercomputer simulations of the coronavirus in recent years, is one such success along with applications in astronomy, fluid dynamics, and other domains. I will illustrate how these application’s successes are based on features of the MOPM.

Charm++ provides a good foundation for development of higher-level languages and frameworks as demonstrated by Adaptive MPI, Charades (discrete event simulation framework), etc.

I will present my assessment of the success and failures of this model over the past two decades, future prospects for it and its software ecosystem, as well as research opportunities.

Biography
Professor Laxmikant Kale is the director of the Parallel Programming Laboratory and Research Professor as well as the Paul and Cynthia Saylor Professor Emeritus of Computer Science at the University of Illinois at Urbana-Champaign.

Prof. Kale has been working on various aspects of parallel computing, with a focus on enhancing performance and productivity via adaptive runtime systems, and with the belief that only interdisciplinary research involving multiple CSE and other applications can bring back well-honed abstractions into Computer Science that will have a long-term impact on the state-of-art.

His collaborations include the widely used Gordon-Bell award winning (SC 2002) biomolecular simulation program NAMD and other collaborations on computational cosmology, quantum chemistry, rocket simulation, space-time meshes, and other unstructured mesh applications.

He takes pride in his group’s success in distributing and supporting software embodying his research ideas, including Charm++, Adaptive MPI and Charm4Py. He and his team won the HPC Challenge award at Supercomputing 2011, for their entry based on Charm++.

Prof. Kale is a fellow of the ACM and IEEE, and a winner of the 2012 IEEE Sidney Fernbach award.

More on Laxmikant Kale: https://charm.cs.illinois.edu/~kale/

More on CESG Seminars: HERE

Please join on Monday, 9/25/23 at 10:20 a.m. in ETB 1034.

Friday, October 6, 2023
10:20 a.m. – 11:10 a.m. (CST)
ETB 1020

Bingzhe Li
Assistant Professor I Dept. of Computer Science
University of Texas at Dallas

Title: “Next-Generation Storage Systems for Big Data”

Talking Points: 

• How to efficiently manage current emerging storage devices
• How to design a new DNA storage system
• How to build a high-performance storage system for big-data applications

Abstract
Tremendous technology developments have been witnessed in the area of computing, network and storage systems. A huge amount of digital data was generated in the past decades with the rapid growth of new technology development such Internet of Things (IoT) devices, edge devices, sensors, 5G, and so forth. Such vast amounts of digital data are being generated and are available for access to all new applications. It becomes a critical and increasing challenge to manage this huge amount of data available at our fingertips and to locate the information that we need at anytime from anywhere.

In this talk, I will focus on emerging storage systems for big data from two perspectives. First, from the capacity perspective, two emerging storage devices/systems with large areal densities (i.e., shingled magnetic recording (SMR) and DNA storage) are introduced including their management and utilization. For SMR, based on its unique properties, we introduce a Machine Learning (ML) based scheme to improve the performance of the SMR storage system. Moreover, for image-based applications, we will introduce how to efficiently store images into DNA storage with higher reliability and capacity. Secondly, from the performance perspective, I will present a high-performance system to accelerate graph neural networks (GNNs). The system is co-designed with the storage systems and algorithms in GNNs and finally can significantly speed up the training process of GNNs.

Biography
Dr. Bingzhe Li is currently an assistant professor of Computer Science at the University of Texas at Dallas. He received his PhD degree in Electrical and Computer Engineering from the University of Minnesota, Twin Cities in 2018 after which he worked as a postdoctoral associate in the Department of Computer Science and Engineering, University of Minnesota, Twin Cities.

His research interests focus on memory and storage systems, computer architecture, and low-cost computing architecture. He has served on conference organization committees as well as technical program committees and as reviewer for several major conferences and journals in computer system, storage, and computing architecture. In recognition of his research, he received the Best Paper Nomination at the ICCD’21 and the featured paper of the Month at IEEE Transactions of Computers on March 2021.

More on Bingzhe Li: Bingzhe Li (tamu.edu)

More on CESG Seminars: HERE

Please join on Friday, 10/6/23 at 10:20 a.m. in ETB 1020.

Friday, September 15, 2023
10:20 a.m. – 11:10 a.m. (CST)
ETB 1020

Shuiwang Ji
Professor at Dept. of Electrical & Computer Engineering
Texas A&M University

Title: “AI for Science in Quantum, Atomistic, and Continuum Systems”

Talking Points: 

• How to use AI to solve problems in physics, chemistry, material science, and fluid mechanics
• How to capture symmetries in physical systems
• How to use AI for advance basic science

Abstract

Advances in AI are fueling a new paradigm of discoveries in natural sciences. Today, AI has started to advance natural sciences by improving, accelerating, and enabling our understanding of natural phenomena at a wide range of spatial and temporal scales, giving rise to a new area of research known as AI for science (AI4Science). In this talk, I will provide an overview of our recent work on understanding the physical world from the subatomic (wavefunctions and electron density), atomic (molecules, proteins, materials, and interactions), to macro (fluids, climate) scales. My talk will focus on how to capture symmetries in physical systems using equivariant models. I will also touch on a few other technical challenges, including explainability, out-of-distribution generalization, and knowledge transfer with foundation and large language models. My talk will be a summary of our recent 263-page review paper available at https://arxiv.org/abs/2307.08423

Biography

Dr. Shuiwang Ji is currently a Professor and Presidential Impact Fellow in the Department of Computer Science & Engineering, Texas A&M University. His research interests include machine learning and AI for science. Dr. Ji received the National Science Foundation CAREER Award in 2014. Currently, he serves as an Associate Editor for IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), ACM Transactions on Knowledge Discovery from Data (TKDD), and ACM Computing Surveys (CSUR). He regularly serves as an Area Chair for ICLR, ICML, and NeurIPS. Dr. Ji is a Fellow of IEEE and AIMBE, and a Distinguished Member of ACM.

Recent Paper: https://arxiv.org/abs/2307.08423

More on Shuiwang Ji: http://people.tamu.edu/~sji/

More on CESG Seminars: HERE

Please join on Friday, 9/15/23 at 10:20 a.m. in ETB 1020.

Friday, September 8, 2023
10:20 a.m. – 11:10 a.m. (CST)
ETB 1020

P. R. Kumar
Professor at Dept. of Electrical & Computer Engineering
Texas A&M University

Title: “Security of Cyber-Physical Systems: Theory and Applications of the Dynamic Watermarking Method”

Talking Points: 

• How to protect our critical infrastructure from cyber-attacks?
• How to secure the power grid, oil refineries, autonomous vehicles?
• What sort of guarantees can be provided?

Abstract

The coming decades will see the large scale deployment of networked cyber–physical systems to address global needs in energy, water, health care, transportation, etc. However, as recent events have shown, such systems are vulnerable to cyber attacks.  We present a general purpose technique, called “dynamic watermarking,” for detecting malicious activity in networked systems of sensors and actuators. It provides a provable guarantee of detection of any non-zero power attack. This method has been implemented in several systems of interest. We present the results of attacks on an autonomous automobile, a grid-tied photovoltaic system, a process control system, and a tethered helicopter. We also present the results of simulation studies of attacks on larger scale systems such as the power grid, through attacks on its automatic gain control loop, and attacks on the Tennessee-Eastman model, an open source benchmark that has been developed for the purpose of evaluating process control technology used in industries such as chemical plants and oil refineries. [Joint work with Kenny Chour, Tong Huang, Hasan Ibrahim, Gopal Kamath, Jaewon Kim, Woo Hyun Ko, Tzu-Hsiang Lin, Jorge Ramos-Ruiz, Bharadwaj Satchidanandan, Lantian Shangguan, Bin Wang, Prasad Enjeti, Swaminathan Gopalswamy, and Le Xie].

Biography

Dr. Kumar has a B.Tech. from IIT Madras (1973), and D.Sc. from Washington University (1977). After serving in the Math Department, UMBC (1977-84), and ECE and CSL, UIUC (1985-2011), he joined Texas A&M. He is a member of the U.S. NAE, the World Academy of Sciences, and the Indian NAE. He was awarded an honorary doctorate by ETH, Zurich. He has received the Alexander Bell Medal of IEEE, the IEEE Field Award for Control Systems, and the Outstanding Contribution Award of ACM SIGMOBILE. He has also received the Eckman Award of AACC, the Ellersick Prize of IEEE COMSOC, the Infocom Achievement Award, and the SIGMOBILE Test-of-Time Paper Award. He is a Fellow of IEEE, an ACM Fellow, and Fellow of IFAC. He was awarded the Distinguished Alumnus Award from IIT Madras, Alumni Achievement Award from Wash U, and Drucker Eminent Faculty Award from CoE, University of Illinois. His current research focus includes reinforcement learning, security, privacy, power systems, automated transportation, unmanned aerial vehicle traffic management, millimeter wave 5G, snd cyber-physical systems.

More on P.R. Kumar: P. R. Kumar (tamu.edu)

More on CESG Seminars: HERE

Please join on Friday, 9/8/23 at 10:20 a.m. in ETB 1020.

Friday, September 1, 2023
10:20 a.m. – 11:10 a.m. (CST)
ETB 1020, In-Person Presentation Only

Sunil Khatri
Professor at Dept. of Electrical & Computer Engineering
Texas A&M University

Title: “Flash – the ‘Overlooked’ Technology”

Talking Points: 

• Flash technology has historically been used for non-volatile memory
• Can flash be used for digital, analog and mixed-signal circuits also?
• This talk posits that it can, and thus holds great promise for VLSI circuit design broadly

Abstract

Flash has been the workhorse technology for non-volatile memories for many years now. In this talk, I show that flash technology can be used to design a variety of general-purpose circuits, both digital and analog. This is demonstrated via case studies that demonstrate two styles of flash-based ASIC design (including a secure variant), flash-based convolutional neural network accelerators (both analog and digital variants), flash-based in-memory computing designs, as well as flash-based analog circuits like DACs and LDOs. Through these studies, we demonstrate several advantages of flash-based designs over conventional CMOS designs,and argue that flash is an overlooked technology in digital and analog design. Some of these advantages of flash are not present in CMOS, such as performance tunability, the ability to counteract circuit aging, the control of speed binning, and the ability to mitigate process variations.

Based on our findings, we posit that the programmability, robustness, stability, and maturity of flash give it a significant edge over the class of “emerging” technologies, making flash a viable technology to eventually replace CMOS.  We hope that our body of work on flash will encourage further research and deployment in the arena of scaling flash to smaller process node geometries, thereby allowing flash to become a key technology for digital and analog circuits in the future.

Biography

Dr. Sunil P Khatri received his PhD from UC Berkeley, his MS from UT Austin, and his BS degree from IIT Kanpur (India). He currently serves as a Professor in ECE at Texas A&M University. His research areas are VLSI IC/SOC design (including hardware-based machine intelligence, secure hardware design, classical and quantum logic synthesis, radiation tolerant design and fast clocking), algorithm acceleration using custom hardware, FPGAs and GPUs, and interdisciplinary extensions of these topics to other areas like communication, DSP, IoT and genomics.

He has co-authored the first papers in many areas, resulting in impactful contributions that changed industrial practice in the areas of regular fabric-based VLSI design approaches, cross-talk canceling CODECs to eliminate on-chip and off-chip crosstalk in VLSI bus interconnect, GPU-based acceleration of VLSI CAD algorithms, and high-speed off-chip output drivers with self-adjusting impedance. Dr. Khatri has over 280 peer reviewed publications. Among these papers, 5 received a best paper award while 7 others received best paper nominations.

More on Sunil Khatri: Sunil Khatri (tamu.edu)

More on CESG Seminars: HERE

Please join on Friday, 9/1/23 at 10:20 p.m. in ETB 1020.