In today’s AI infrastructure, traditional copper and pluggable optics are ineffective in scaling package-level compute advancements to the system rack and row levels, leading to low efficiency, high power consumption, and high costs. New technologies are needed to support growing model sizes and complexity. Ayar Labs' in-package optical I/O solution enables peak platform performance by providing efficient, low-cost scaling at the rack and row levels. It also offers extended accelerator memory to optimize the balance between memory and compute. In this presentation, Mark Wade will show application-level improvements in performance and TCO metrics, such as productivity, profitability, and interactivity, using optical I/O-based scale-up fabrics for inference and training.
Mark Wade
Mark is the Chief Executive Officer and Co-Founder of Ayar Labs. His prior roles at Ayar Labs include Chief Technology Officer and Senior Vice President of Engineering. He is recognized as a pioneer in photonics technologies and, before founding the company, led the team that designed the optics in the world's first processor to communicate using light. He and his co-founders invented breakthrough technology at MIT and UC Berkeley from 2010-2015, which led to the formation of Ayar Labs. He holds a PhD from the University of Colorado.
Ayar Labs
Website: https://ayarlabs.com/
Ayar Labs is transforming AI infrastructure by accelerating data movement. Recognizing that the complexity and size of AI models are increasing at a rate that traditional interconnect technology cannot handle, the company has developed the industry’s first optical I/O solution that enables customers to maximize the compute efficiency and performance of growing AI infrastructure, while reducing costs, latency and power consumption. Based on open standards and optimized for both AI training and inference, Ayar Labs’ optical I/O solution is backed by a robust ecosystem that enables it to integrate smoothly into AI systems at scale.