Ayar Labs to Accelerate Development, Application of Optical Interconnects in AI/Machine Learning Architectures with NVIDIA


Reading time ( words)

Ayar Labs, the leader in chip-to-chip optical connectivity, is developing with NVIDIA groundbreaking artificial intelligence (AI) infrastructure based on optical I/O technology to meet future demands of AI and high performance computing (HPC) workloads. The collaboration will focus on integrating Ayar Labs’ technology to develop scale-out architectures enabled by high-bandwidth, low-latency and ultra-low-power optical-based interconnects for future NVIDIA products. Together, the companies plan to accelerate the development and adoption of optical I/O technology to support the explosive growth of AI and machine learning (ML) applications and data volumes.

Meeting Future Performance and Power Requirements with Optical I/O

Optical I/O uniquely changes the performance and power trajectories of system designs by enabling compute, memory and networking ASICs to communicate with dramatically increased bandwidth, at lower latency, over longer distances and at a fraction of the power of existing electrical I/O solutions. The technology is also foundational to enabling emerging heterogeneous compute systems, disaggregated/pooled designs, and unified memory architectures that are critical to accelerating future data center innovation.

“Today’s state-of-the-art AI/ML training architectures are limited by current copper-based compute-to-compute interconnects to build scale-out systems for tomorrow’s requirements,” said Charles Wuischpard, CEO of Ayar Labs. “Our work with NVIDIA to develop next-generation solutions based on optical I/O provides the foundation for the next leap in AI capabilities to address the world’s most sophisticated problems.”

Delivering the ‘Next Million-X’ Speedup for AI with Optical Interconnect

“Over the past decade, NVIDIA-accelerated computing has delivered a million-X speedup in AI,” said Rob Ober, Chief Platform Architect for Data Center Products at NVIDIA. “The next million-X will require new, advanced technologies like optical I/O to support the bandwidth, power and scale requirements of future AI and ML workloads and system architectures.” 

As AI model sizes continue to grow, by 2023 NVIDIA believes that models will have 100 trillion or more connections – a 600X increase from 2021 – exceeding the technical capabilities of existing platforms. Traditional electrical-based interconnects will reach their bandwidth limits, driving lower application performance, higher latency and increased power consumption. New interconnect solutions and system architectures are needed to address the scale, performance and power demands of the next generation of AI. Ayar Labs’ collaboration with NVIDIA is focused on addressing these future challenges by developing next-generation architectures with optical I/O.

Share




Suggested Items

CES 2021 Coverage: A Virtual Show Floor

01/17/2021 | Dan Feinberg, Technology Editor, I-Connect007
I write this on the final day of CES 2021, and I expect CES will never be the same. It will not revert back to what it once was. I also cannot imagine it stays a totally virtual show; in doing so, I feel it would fail. Does that mean I think the 2021 show was a failure? No, not at all. In fact, it was a very good event, particularly in light of the medical and political pandemic that we have been enduring.

Liquid Cooling Moves Onto the Chip for Denser Electronics

10/05/2015 | Georgia Institute of Technology
Using microfluidic passages cut directly into the backsides of production field-programmable gate array (FPGA) devices, Georgia Institute of Technology researchers are putting liquid cooling right where it's needed the most - a few hundred microns away from where the transistors are operating.



Copyright © 2022 I-Connect007. All rights reserved.