At the Nvidia GPU Technology Conference (GTC) 2021, he did The company provided A new system on a chip for self-driving cars. Its name is Drive Atlan and it is a next-generation GPU architecture that will be implemented in 2025 models.
Likewise, Drive Atlan comes with new ARM CPU cores. Plus BlueField’s data processing unit, which will help complete chip workloads. Specifically, processes related to artificial intelligence are essential to enable automated systems of cars. In this regard, Nvidia CEO Jen-Hsun Huang expressed:
Our new Atlan is truly a technology marvel, integrating all of Nvidia’s strengths in AI, automotive, robotics, security, and secure data centers at BlueField to deliver safe and autonomous fleets. The transportation industry needs a computing platform it can rely on for decades. The software is too bulky to be replicated in every car. “
Drive Atlan will be able to perform more than 1000 billion operations per second
Atlan technology will allow it to perform more than 1,000 trillion operations per second. Three times more than Orin can do, the chip that will arrive with the company’s cars in 2022.
In fact, the 2025 models will come with more power and potential for improvement through software updates via OTA (Over The Air, for English). To do this, Atlan will use the same CUDA and TensorRT APIs and libraries it uses on existing Drive slides.
Volvo AURIN will be used in its upcoming vehicles
Among other announcements, Nvidia said Drive Orin will be used in upcoming cars of the Swedish brand Volvo. In fact, this will be presented next year in the XC90 model. Its job will be to ensure that the LiDAR and Vision sensor data is processed.
This function complements the vehicle’s basic operations, namely, energy management and driver assistance. Together they will ensure the good running of the vehicle.
Nvidia introduced a new Hyperion platform
In addition to Drive Atlas, the company introduced Drive Hyperion 8. An update targeting the “AI supercomputer on wheels” that drives the development of Tier 4 autonomous or high-automation vehicles.
This includes two SoC DRIVE Orin, which are responsible for processing “data from 12 external cameras, three internal cameras, nine radars and a LiDAR sensor”.
With these announcements, Nvidia demonstrates its intent to improve its vehicle’s driving system, while easing the path to autonomy. For now, we can’t help but wait to see how much they improve their cars.
Great Wall is planning to launch a hydrogen-powered vehicle