A performance car isn’t judged on horsepower alone. Acceleration, braking, cornering speed and suspension are among the features that define how a car will fare on the road.
The same is true for the processors inside an autonomous vehicle. While processing speed (measured in trillions of operations per second, or TOPS) is a key performance indicator, it’s the architecture, the programmability and the AI software stack running that determine excellence.
Today, NVIDIA is ushering in a new era of AI computing and software for autonomous driving.
At the GPU Technology Conference, we introduced NVIDIA DRIVE AGX Orin, a highly advanced software-defined platform for autonomous vehicles. The platform is powered by a new system-on-a-chip (SoC) called Orin, which consists of 17 billion transistors and is the result of four years of R&D investment. It’s an SoC born out of the data center.
Orin achieves 200 TOPS — nearly 7x the performance of the previous generation SoC Xaiver — and is designed to handle the large number of applications and deep neural networks that run simultaneously in autonomous vehicles and robots, while achieving systematic safety standards such as ISO 26262 ASIL-D.
Orin is software compatible with Xavier, allowing customers to leverage their existing development investments. It’s also scalable — with a range of configurations able to deliver Level 2+ AI-assisted driving up to Level 5 fully driverless operation.
Taking the Pedal to the Metal on Inference
Cars and trucks of the future will require an optimized AI architecture not only for autonomous driving, but also for intelligent vehicle features like speech and driver monitoring. We’ll converse with the car and it’ll respond: answering questions, directing us to destinations, warning us of road conditions. It will be our co-pilot and guardian, able to take over and drive autonomously, monitor our alertness and safeguard us.
Xavier, our current-generation SoC, has proven to be the best-performing processor for inference on the market. In this year’s MLPerf assessment, it ranked as the highest performer under edge-focused scenarios among commercially available edge and mobile SoCs.
Leveraging this industry-topping architecture, Orin will continue to lead the way for edge computing. It’s designed specifically to handle the large number of applications and DNNs running necessary for safe autonomous vehicle operation.
With the introduction of DRIVE AGX Orin, we’re dramatically raising the bar for vehicles starting production in 2022. Combining our industry-leading unified architecture and the open DRIVE Software stack, NVIDIA is redefining vehicle performance for the age of autonomy.