Title: Keeping Pace: Micro-Models, Large Language Models, and the Co-Evolution of Chip and Data Center Design in a Rapidly Shifting AI Landscape Abstract: The field of Artificial Intelligence is experiencing a period of exceptional dynamism, with significant advancements occurring at a rapid pace – often within a timeframe of four to six months. This keynote […]
Title: Universal Memory Interface (UMI): A practical solution to break the memory wall for AI ASICs Abstract: Compute performance demand has been growing exponentially in recent years, and with the advent of Generative AI, this demand is growing even faster. Moore’s law coming to an end as well as the Memory Wall (memory bandwidth & […]
Title: Connectivity for AI Everywhere: The Role of Chiplets Abstract: Chiplet technology is set to be a transformational paradigm. It offers compelling advantages in cost, time-to-market, and power consumption. AI has emerged as a primary catalyst for this trend because the only way to meet the accelerated demands of AI is by using chiplets. By […]
Title: Powering Llama 3: A Sneak-peek into Meta's Massive Infrastructure for Generative AI Abstract: We trained a model. It used a whole lot of GPUs and a giant network. We wrote a paper about it. People thought it was cool. I'll talk about that. Oh, yeah, we open-sourced the model too! And here I asked Llama 3 […]
Title: Can Interconnects Keep Up with AI? YES! Abstract: The bandwidth requirements of AI clusters represent an unprecedented challenge to improve the cost, power, and reliability of AI Fabric interconnects. I will discuss the latest directions to solve these pressing issues. Speaker Biography: Andreas "Andy" Bechtolsheim is Co-Founder and Chief Architect of Arista Networks. He […]
Title: Standardizing Interconnect Test and Repair: Motivation and On-Going Effort Abstract: Chip-let based design, fueled by various advanced packaging technologies, is projected to revolutionize the semiconductor industry. This brings along with it associated challenges to achieve adequate yield and contain aging related reliability issues. This talk will present manufacturing defect profiles associated with advanced packaging; […]
Presenter Names: Priyank Shukla and Ruben Sousa (Synopsys) Abstracts: This tutorial delves into the world of 112G/224G linear electro-optical links, a rapidly growing technology for next-generation PCIe and Ethernet standards. We'll explore the motivations behind this shift, providing a clear understanding of the benefits of linear electro-optical interfaces. The tutorial dives into the ongoing standardization […]
Presenter Names: Dhabaleswar K. (DK) Panda and Hari Subramoni Abstract: High-Performance Networking technologies are generating a lot of excitement towards building next generation HighEnd Computing (HEC) systems for HPC and AI with GPGPUs, accelerators, and Data Center Processing Units (DPUs), and a variety of application workloads. This tutorial will provide an overview of these emerging technologies, […]
Presenter Name: Rajalaxmi Angadi, Juee Himalbhai Desai, and Peinan Zhang (Intel) Abstract: With increase in AI application and optimizations in HPC/AI, there are multiple middleware libraries using libfabric. This tutorial will go through libfabric overview, it will explain the software stack hierarchy. There will be examples programs with point to point and collectives coming from […]
