Arm Research Summit 2019: Catch Up and Highlights
The fourth annual Arm Research Summit, and the first in the US, held in Austin, Texas, created a collaborative space for academics, researchers and industry experts globally to discuss complex computing challenges in the future landscape.
The fourth annual Arm Research Summit, and the first in the US, held in Austin, Texas, created a collaborative space for academics, researchers and industry experts globally to discuss complex computing challenges in the future landscape. The 15-18 September was filled with discussion and networking, enabling delegates to discover new developments, opportunities and connections.
What You Missed
The keynote speakers kicked off each day, each talking about their respective interests and research areas. Topics of computer architecture and post-Moore's law, the Internet of Things, and IT disruptions were discussed, giving each day of the Summit a unique opening. As well as these inspiring keynote speakers and plenty of other presentations from experts in their research fields, the Summit hosted a poster session to give even more delegates the opportunity to present their work. The poster session on the second day saw Amna Shahab win the Best Poster award, her poster titled Exploring New Directions for Reducing Neural Model Training Time. The quality of Amna's work shines through, as she was also awarded 'Best Poster' at the 2018 Summit in Cambridge - we look forward to seeing more exciting research at future events!
Best Presenter was awarded to Tushar Krishna, who spoke in the ML Hardware and Applications session with Enabling Continuous Learning through Neural Network Evolution in Hardware. Jesse Beu from Arm Research, who was poster chair as well as session chair for Tushar's session, gave some insight into why both Amna and Tushar were crowned best in their respective categories:
“What I found especially interesting about both the Best Poster and the Best Presenter winners is that both embraced and reinforced a theme seen throughout the Summit, that as researchers we need to look beyond our abstraction layer and pursue cross-cutting research to maximize impact. Amna’s recognition of the link between hardware efficiency and statistical efficiency for ML training really resonated with attendees because this kind of insight only comes from understanding workloads in addition to the hardware they run on. Similarly, Tushar’s presentation was popular because it pushed the bounds on what it means to be an architect; it showed we shouldn’t limit ourselves to understanding just hardware tradeoffs, we should also look to how the applications plays off these tradeoffs.”
Congratulations to both Amna and Tushar!
In the spirit of collaboration, we've made the presentations across the four days available below. You can also watch recordings of keynote talks, including Rich Wolski, Matthew Scholl, and an overview of Arm Research given by Arm Fellow Stuart Biles.
Day 1
|
Workshop: Impact of Arm Hardware from an HPC Application Perspective (Present and Future-Looking) - Slides |
Day 2
|
Plenary: Stuart Biles: An Introduction to Arm Research - Video | Slides |
|
|
Keynote: Kunle Olukotun: Computer Architecture in the Post Moore's Law Era* |
|
|
HPC Applications I Mohammed Zubair: Optimization of a Solver for Computional Materials and Structures Problems on Arm Processor - Slides Rached Abdelkhalak: High Performance Stencil Computations for Oil and Gas Applications on Arm Architectures* Brian Gravelle: Porting HPC Miniapps to Arm - Slides Shinji Sumimoto: Performance Evaluation of SVE Enabled Arm Processor A64FX using Variable Vector Length - Slides |
Biology and Compute Saugata Ghose: Accelerating Genome Analysis: A Primer on an Ongoing Journey - Slides Andrew Ellington: Yeastons: Implementing Artificial Neural Networks with Engineered Living Cells - Slides Marko Vasic: CRN++: Molecular Programming Language - Slides Renee St. Amant: AI, Ethics, and Policy Considerations when Technology Meets Biology - Slides |
|
System Security Tyler Hunt: System Security: General Challenges and a Cloud GPU System - Slides Simha Sethumadhavan: Memory Safety: To Blacklist or Whitelist? - Slides Shijia Wei: Using Power-Anomalies to Counter Evasive Micro-Architectural Attacks in Embedded Systems - Slides Dongdong She: Efficient Fuzzing with Neural Networks - Slides
|
Workshop: Renegotiating the Levels of Abstraction for the Post Moore's Law Era Matt Horsnell: Abstractions and Specialization - Slides Sarita Adve: Interface, Communication, Approximation - Slides Vikram Adve: Uniform Abstractions for Heterogeneous Parallel Systems - Slides Arrvindh Shriraman: Hardware Construction by Software - Slides Thierry Moreau: The Past, Present, and Future of Deep Learning Acceleration Stacks - Slides Riyadh Baghdadi: Compiling Dense and Sparse Neural Networks using Tiramisu - Slides Jeff Setter: Halide to Hardware - Slides Naums Mogers: Functional Interface for Performance Portability on Parallel Accelerators - Slides Aviral Shrivastava: dMAzeRunner - Accelerating Loop Nests on Dataflow Accelerators - Slides Ramana Radhakrishnan: Arm's Machine Learning Processor - Slides |
|
HPC on Arm Simon McIntosh-Smith: Arm in HPC: from Prototype to Production - Slides Andy Warner: Further Updates from the Coalface of Arm in HPC - Slides JaeHyuk Kwak and Vitali Morozov: Early Experience of the Arm-based High-Performance Computing Eco-system - Slides Kevin Pedretti: Experiences Scaling a Production Arm Supercomputer to Petaflops and Beyond - Slides |
ML Hardware and Applications Jiang Hu: Machine Learning Techniques for Chip Design Verification and Prediction - Slides Muhammad Shahbaz: Taurus: An Intelligent Data Plane - Slides Tianmu Li: 3PXNet: Pruned-Permuted-Packed XNOR Networks for Edge Machine Learning - Slides Tushar Krishna: Enabling Continuous Learning through Neural Network Evolution in Hardware - Slides
|
|
Side Channels and Exploits Jakub Szefer: Three-Step Models for Timing Channels in Processor Caches and TLBs - Slides Caroline Trippel: CheckMate: Automated Exploit Program Generation for Hardware Security Verification - Slides Hovav Shacham: Data Dependent Instruction Timing Channels - Slides Mohit Tiwari: Cyclone: Detecting Contention-Based Cache Information Leaks Through Cyclic Interference - Slides |
HPC Applications II David Donofrio: Embedding HPC at the Edge - Slides Aaron Walden: An Optimized Multicolor Point-implicit Solver for Unstructured Grids on the ThunderX2 Processor - Slides Dirk Pleiter: Enabling HPC Applications for SVE - Slides Oliver Sheridan-Methven: Numerical Simulations using Approximate Random Numbers: VLA Algorithms for SVE and FP16 Capable Hardware - Slides |
|
Memory Systems I - Specialization and Persistency Charles Eckert: Neural Cache: Bit-Serial In-Cache Acceleration of Deep Neural Networks - Slides Daichi Fujiki: Duality Cache for Data Parallel Acceleration - Slides Mohammad Alshboul: Lazy Persistency: A High-Performing and Write-Efficient Software Persistency Technique - Slides Samira Khan: Rethinking System Support for Persistent Memory - Slides |
Emerging Applications Yuhao Zhu: The Next Quintillion Pixels and Beyond: Architecting Next-Generation Mobile Visual Computing Systems - Slides Ramesh Govindan: Augmented Vehicular Reality - Slides Arslan Munir: Design and Evaluation of MPSoC ECU Architectures for Secure and Dependable Automotive Cyber-Physical Systems - Slides Dayane Reis: A Computing-in-Memory for Searching on Homomorphically Encrypted Data - Slides |
|
Security Meetup: Slides |
|
Day 3
|
Keynote: Rich Wolski: Devices-as-Services and The Internet as a Platform of Things - Video | Slides |
|
|
Panel: Rethinking Boundaries through Hardware-Software Co-design for Productive Post-Moore Computing - Slides |
|
|
New Technologies in Circuits and Systems Sung Kyu Lim: Heterogeneous 3D ICs: Benefits, Challenges, and Future Prospects - Slides Haitong Li: On-Chip Memory Technology Design Explorations for Mobile Deep Neural Network Accelerators - Slides Aseem Sayal: All-Digital Time-Domain CNN Engine Using Bi-Directional Memory Delay Lines for Energy Efficient Edge Computing* Azad Naeemi: Performance Modeling and Design for Various Spintronic Memory Technologies - Slides |
System Architecture Benjamin Lee: The Computational Sprinting Game - Slides Mattan Erez: QoS for Accelerator-Rich "Fat" Nodes - Slides Saptadeep Pal: Waferscale Processors - Slides Ronald Dreslinski: Introspective Computers - Slides |
|
Modeling David C: Enzian: An Open Research Computer, In Stores Now - Slides Matthew Baker: Scaling Full-system Simulation of Arm SVE Processors Using Compilers and Runtime Tool APIs - Slides Yuetsu Kodama: RIKEN Fugaku Processor Simulator and its Accuracy for Cache and Memory Throughput - Slides Ben Simner: Systems-level Architecture Modelling - Slides |
Workshop: Your Research Enabled by Arm - SoC Labs John Goodenough: Executive Introduction* Plout Galatsopoulos: Arm Flexible Access for Research & Arm Research SoC Labs - Slides Kobus Marneweck: DesignStart Cortex A5 subsystem - Slides John Goodenough: Arm Education Plans* Paul Whatmough: Harvard SoC designs - Slides Liam Dillon: AFA Technical Offerings - Slides Liam Dillon: Why Bother with Debug - Slides Mark Willoughby: STFC Europractice and Arm SoC Labs - Slides Andrew Kahng: The OpenROAD project - Slides Tutu Ajayi: Arm Coherent Accelerator Interface - Slides |
|
Workshop: Arm HPC User Group - Slides |
|
|
IoT Technology Joshua San Miguel: Architecting for Intermittence - Slides Filip Maksimovic: Power, Ground, and Antenna: Three-Pin Chips for Standards Compatible Wireless Networks* Irina Alam: Fault Resilient Voltage Scaling of Embedded Memories for Efficient Inference or Operation at the Edge - Slides I-Ling Yen: Semantic based Data Management and Discovery in the IoT-Edge-Cloud Infrastructure - Slides |
Memory Systems II: Coherence and Prefetching Sarita Adve: Dynamic Coherence Specialization - Slides Hao Wu: Temporal Prefetching without the Off-chip Metadata - Slides Paul Gratz and Daniel Jimenez: Speculative, Coordinated Memory Hierarchy Management - Slides Tony Nowatzki: Stream-based Memory Access Specialization for General Purpose Processors - Slides |
|
Benchmarks and Software Vijay Janapa Reddi: The Vision Behind MLPerf (mlperf.org): A Community-driven Benchmark Suite for ML Frameworks, ML Accelerators and ML Systems in Cloud and Edge Computing - Slides Patrick Lavin: Using the Spatter Benchmark Suite to Evaluate SVE Support for Gather/Scatter - Slides Muhammad Huzaifa: Enabling Architecture Research for Augmented and Virtual Reality - Slides Jonathan Helmus: Creating a Python Data Science Distribution for Arm with Conda - Slides |
Cloud, Edge and IoT Jan S. Rellermeyer: Kubernetes on the Edge - A Peer-to-Peer Approach - Slides Babak Zamirai: SIEVE: Speculative Inference on the Edge with Versatile Exportation - Slides Kamyar Mirzazad Barijough: Distributed Deep Learning Inference On Resource-Constrained IoT Edge Clusters - Slides Jayashree Mohan: CrashMonkey and Ace : Systematically Testing File-System Crash Consistency - Slides
|
|
Cross-stack Optimizations Vikram Adve: ApproxHPVM: Accuracy-aware Optimizations for Heterogeneous System-on-Chip Processors - Slides Simone Campanoni: Liberating Threads from Non-Numerical Programs with an Architecture-Compiler Co-Design - Slides Antonino Tumeo: Data and Model Convergence: a Case for Software Defined Architectures - Slides Timothy Rogers: Leveraging Cross-Stack Solutions to Improve Massively Parallel Systems* |
Memory Systems III: Processing in or near Memory Mochamad Asri: The Non-Uniform Compute Device (NUCD) System Architecture - Slides Miquel Moreto: Hardware Acceleration of Sparse Data Rearrangement Near Memory - Slides Sriseshan Srikanth: Strider: Architectures for Scalable Memory Centric Reduction of Sparse Data Streams - Slides Jiayi Huang: Active-Routing: Compute on the Way through Near-Data Processing - Slides |
Day 4
|
Keynote: Matthew Scholl: The Upcoming Disruption in IT and what NIST Looks to Prioritize for Cybersecurity - Video | Slides |
|
|
Panel: A Conversation with the NSF - Partnerships and Collaborations - Slides |
|
|
Microcontroller Security and Applications Joseph Yiu: Evolution of Security Solutions for Microcontrollers - Slides Reza Azarderakhsh: Post-Quantum Cryptography in Arm Cortex M Series - Slides Jeongsup Lee: A Self-tuning Cortex-M0 IoT Processor Based on Leakage-Ratio Measurement for Energy-Optimal Operation - Slides Jiecao Yu: Efficient Low-Precision Deep Neural Networks on IoT Microcontrollers* |
Accelerators James C. Hoe: Do more with FPGAs in Computing: from Field Programmable to Programmable - Slides Arrvindh Shriraman: uIR: An Intermediate Representation for Constructing Hardware Accelerators - Slides Thierry Moreau: A Hardware-Software Blueprint for Flexible Deep Learning Specialization - Slides Christopher Batten: Architectural Specialization for Dynamic Task-Parallel Programs - Slides |
|
ML Algorithms Hadi Esmaeilzadeh: AI for Optimized AI Execution - Slides Alex Beatson: Accelerating Automated Modeling and Design with Stochastic Optimization, Probabilistic Models, and Neural Networks - Slides Saavan Patel: Hardware-Aware, Scalable, Combinatorial Optimization on a Boltzmann Machine - Slides Mohsen Imani: A Framework for Secure Collaborative Learning* |
|
|
ML Meetup: Slides Novel Algorithm Update: Slides |
|
* The slides for these sessions are unavailable at the speakers' request.
Arm Research Summit 2020
We are excited to announce that the Arm Research Summit is returning for 2020 and will be held in Europe! Further details will be released soon, so keep an eye out so you can be the first to submit your work and register for the fifth annual Summit!
Re-use is only permitted for informational and non-commercial or personal use only.
