Linley Fall Processor Conference 2020

Held October 20-22 and 27-29, 2020
Proceedings available

» Events  |  Event Info  |  Agenda Overview  |  Day One  |  Day Two  |  Day Four  |  Day Five  |  Day Six  |  Proceedings & Event Videos

Agenda for Day Three: Thursday October 22, 2020
View Day One

Click arrow below to view presentation abstract or click here -> to view all abstracts
8:30am-10:00amSession 4: Automotive Processor Design

Next-generation ADAS equipment creates a path to eventual deployment of autonomous vehicles, but these systems require high-performance AI acceleration that meets automotive power and safety requirements. These systems focus on computer vision, often analyzing inputs from multiple camera simultaneously. This session, moderated by The Linley Group senior analyst Mike Demler, explores the challenges in advanced automotive designs and the technologies that are becoming available to meet them.

Domain-Specific Architectures Enable Highly Efficient Deep Learning at the Edge
Avi Baum, CTO, Hailo Technologies

Decades of domain-agnostic computer architectures have reached asymptotic performance. Major improvements can only come through disruption. Effective architectural disruption requires extensive domain understanding. Deep learning for vision is a domain with growing traction that requires high compute capacity for edge processing. The talk will cover some of the challenges and pitfalls with this approach while highlighting programming-model related implications inherent to this concept.

Scalable, High-Performance Processing for Next-Generation Autonomous Systems
Srikanth Rengarajan, Manager Automotive IP, North American Partners, Arm and Jinson Koppanalil, Distinguished Engineer, Arm

Development of autonomous systems continues to accelerate across multiple vertical segments. This trend creates a challenging transition to consolidated systems in industrial and domain controllers within automotive, driven by the promise of standardized platforms with robust functional safety and security. Co-locating processing contexts with such divergent requirements requires flexible designs that scale with application needs. Next generation processors must provide high performance while meeting stringent functional safety requirements. This presentation explores new technologies that are evolving to enable these characteristics.

Architecting Safe and Secure Solutions for Autonomous Vehicles
Fergus Casey, R&D Director, Synopsys

Computer vision can eliminate human error and make vehicles safer – effectively removing driver uncertainty. Autonomous vehicles are expected to save almost 300,000 lives per decade in the United States alone. But despite decades of concepts and development, ubiquitous “driverless” vehicles seem a distant reality. This presentation will explore the challenges that SoC designers and OEMs face when developing autonomous vehicles and will describe key milestones and metrics that the industry, and SoC designs, must achieve to enable autonomous driving.

There will be Q&A and a panel discussion featuring the above speakers.

10:00am-10:10amBreak Sponsored by Arm
10:10am-11:40amSession 5: Security

End-to-end security solutions are becoming increasingly important as hackers are exploiting hardware that exists in every facet of our lives. These exploitations target different kinds of silicon ranging from cloud solutions to low-power MCUs. This session, moderated by The Linley Group senior analyst Aakash Jani, discusses a scalable cloud solution for hardware security modules, a malicious firmware detection system for MCUs, and a data fabric that implements fully homomorphic encryption.

Breaking Out of the HSM Box
Ben Levine, Head of Product, Security Solutions Business Unit, Marvell

Hardware Security Modules (HSMs) have existed for almost 50 years. A typical HSM is a physical box located in a business or on a private LAN that stores and protects keys and performs cryptographic operations. But as businesses move off-premises and into the cloud, this model is no longer viable. This presentation will explore a new paradigm that provides businesses with flexible, scalable, distributed, high-performance HSMs designed specifically for the cloud and lights-out datacenters without compromising on security.

End-to-End Supply Chain Protection with Dynamic Trust
Srirama (Shyam) Chandra, Security System Architect, Lattice Semiconductor

The security paradigm is changing; firmware attacks are a growing and serious concern across all end markets. TPM and MCU-based "static" solutions are not sufficient for comprehensive protection needed in a constantly changing and increasingly risky supply chain environment. This presentation will discuss how Lattice Sentry products address these challenges by offering an FPGA-based parallel real-time “dynamic” solution, based on the NIST 800-193 platform firmware resiliency (PFR) guidelines, to protect, detect, and recover from unauthorized firmware changes.

Next-Generation Machine-Learning Algorithms for Privacy-Preserving Neural Networks
Paul Master, CTO and cofounder, Cornami

When applied to machine learning, the new Internet security standard, Fully Homomorphic Encryption (FHE), provides a game-changing set of features to ML algorithms, paving the way to ubiquitous adoption. Conventional processing chips cannot provide real-time FHE performance due to the high computational and data-movement costs inherent in the algorithm itself. This presentation will describe the unique TruStream Computational Fabric, which uses both temporal and spatial properties to enable deployment of real-time FHE applications.

For this session, each talk will have 10 minutes of Q&A immediately following.

11:40am-12:40pmBreakout sessions with today's speakers
1:30pm-3:30pmSpeaker 1:1 Meetings

 

Premier Sponsor

Platinum Sponsor

Gold Sponsor

Andes Technologies

GSI Technology

Industry Sponsor

Media Sponsor