Track 3 Late Morning
Domain Specific Accelerator (DSA) Architectures for Signal Processing, Communications, and Machine Learning
Abstract
This tutorial will explore Domain Specific Accelerator (DSA) architectures, which are specialized hardware accelerators designed for specific application domains such as signal processing, communications, and machine learning. As the demand for high-performance and energy-efficient systems grows, DSAs offer a promising solution by providing tailored hardware solutions that outperform general-purpose processors in specific tasks. The tutorial will cover the fundamental principles of DSA design, including architecture, optimization techniques, and practical applications. Special emphasis will be placed on the latest advancements in AI, generative AI, and Large Language Models (LLMs). Attendees will gain insights into the design and implementation of these accelerators for various applications.
Speaker
Dr. Kiran Gunnam
Dr. Kiran Gunnam is an innovative technology leader with vision and passion who effectively connects with individuals and groups. His breakthrough contributions are in the areas of advanced error correction systems, storage-class memory systems, and computer vision-based localization & navigation systems. He has helped drive organizations to become industry leaders through ground-breaking technologies. He has 86 issued US patents and 100+ patent applications/invention disclosures on algorithms, architectures, and real-time low-cost implementations for computing, storage, computer vision, and AI systems. He is the lead inventor/sole inventor for 90% of them. His patented work has been incorporated in more than 3 billion data storage, Wi-Fi, and 5G chips as of 2020 and is set to continue to be incorporated in more than 500 million chips per year. Dr. Gunnam is also a key contributor to the precise localization and navigation technology commercialized for autonomous aerial refueling and space docking applications. His recent patent-pending inventions on low-complexity simultaneous localization and mapping (SLAM) and 3D convolutional neural network (CNN) for object detection, tracking, and classification are commercialized for LiDAR+ camera-based perception for autonomous driving and robotic systems. His more recent inventions on machine learning accelerators have ~2x savings vs the state of the art.
Dr. Gunnam has been involved with the IEEE standards association (SA) since 2013. He is a member of IEEE Computer Society’s Microprocessors Standards Committee and is the Chair of IEEE P3109 Standards Working Group for Arithmetic for Machine Learning. He is also the Chair of IEEE CASS Standard Activities Subdivision (SASD). He is also a member of the Board of Governors of the IEEE Circuits and Systems Society (CASS) for 2021-2022. Dr. Gunnam served as IEEE Distinguished Speaker and Plenary Speaker for 30+ events and international conferences, and more than 4000 attendees benefited from his talks. Dr. Gunnam also served as a lead Instructor for machine learning and deep learning workshops organized by ACM in collaboration with IEEE and ValleyML from 2018 to 2020. Dr. Kiran Gunnam is a recipient of the ValleyML Distinguished Technical Achievement Award for long-lasting contributions to architectures and algorithms of real-time signal processing, communication, and machine learning systems that enabled ubiquitous computing.
