Recogni Is Making AI-Based Vision Inference Chips

Recogni is an AI chip start-up incorporated in 2017 for the automotive market. Like most companies, the pandemic threw a curve at them. After only four months into their design effort, everyone had to go home. Things worked out for the better for Recogni, though, as in the span of 14 months, they had their first fully functional 7nm chip using Cadence full flow in the cloud. Today, they build the world’s highest-performing chips that are fully functional for vision inference in autonomous vehicles. A key role in achieving this was Recogni’s partnership with Cadence and Google Cloud.

Recogni’s vision inference silicon has a combination of power and performance that’s ideal for the automotive market. It’s crucial to process all the data coming in from the vehicles and simultaneously burn as little energy as possible without impacting the vehicle range. For designing complex chips in the 5nm to 7nm range that they were aiming for, Recogni turned to Cadence’s state-of-the-art cloud-based tools via the Cadence Cloud Passport Model.

Design, verification, synthesis, and most importantly, early place and route—getting all the elements right on the first try is critical, especially for a start-up like Recogni. For them, Cadence’s tools have been very versatile in all of those domains, and the ability to run Cadence full flow in Google Cloud was essential to Recogni’s success. The ability of Cadence tools to seamlessly scale across multi-core Google Cloud instances proved to make a big difference

“Designed with Cadence” is a series of videos that showcases creative products and technologies that are accelerating industry innovation using Cadence tools and solutions. Learn more about how Recogni is building the world’s highest-performing AI chips.

For more Designed with Cadence videos, check out the Cadence website and YouTube channel.

Learn more about the Cadence Cloud Portfolio.