From Cloud to Car: How NIO Develops Intelligent Vehicles on NVIDIA HGX

Making future-generation smart vehicles calls for an AI infrastructure that pushes the slicing edge.

Electrical car maker NIO is using NVIDIA HGX to build a extensive facts centre infrastructure for developing AI-driven, application-described automobiles. With substantial-general performance compute, the automaker can continuously iterate on innovative deep understanding models, producing robust autonomous driving algorithms in a closed-loop atmosphere.

“The sophisticated situations confronted by mass-made cars and trucks and the huge volume of data these fleets produce are the cornerstones of NIO’s autonomous driving capabilities,” explained Bai Yuli, head of AI Platforms at NIO. “By making use of NVIDIA high-performance compute remedies, NIO can speed up the route to autonomous driving.”

NIO has already launched clever autos formulated on this infrastructure, these as its entirely electric, intelligent flagship sedan, the ET7. Its mid-dimensions overall performance sedan, the ET5 is scheduled to debut in September.

In addition to superior-functionality details middle improvement, equally products are crafted on the Adam supercomputer, powered by four NVIDIA Drive Orin methods-on-a-chip. These vehicles aspect autonomous driving and clever cockpit capabilities that are consistently iterated upon and improved in the knowledge center for a revolutionary consumer experience.

Building a High-Efficiency AI Infrastructure With NVIDIA GPUs and Networking

The purpose of the details center is to ingest, curate and label enormous amounts of knowledge for AI design education at scale.

Data selection fleets create hundreds of petabytes of facts and billions of illustrations or photos every year. This information is then utilised to enhance the deep neural networks (DNNs) that will run in the vehicles.

NIO’s scalable AI infrastructure is powered by NVIDIA HGX with 8 A100 Tensor Main GPUs and NVIDIA ConnectX-six InfiniBand adapters. This scalable supercomputer cluster is made up of NVME SSD servers and interconnects by way of the higher-speed NVIDIA Quantum InfiniBand community platform.

This highly effective infrastructure permits substantial quantities of deep understanding coaching details to be transferred to supercomputer memory or NVIDIA A100 video memory at extremely-substantial speeds of up to 200Gbps.

NVIDIA HGX A100 is a higher-effectiveness server system intended for AI situations, which includes big datasets and challenging products like people that ability autonomous vehicles. It incorporates a absolutely optimized NVIDIA AI application stack in NGC.

The platform sets a new compute density benchmark, condensing 5 petaflops of AI functionality and replaces siloed infrastructures with a single platform for a wide selection of elaborate AI apps.

With the HGX A100, NIO is ready to flexibly produce and deploy scalable AI techniques. It also permits the corporation to raise model advancement performance by up to 20x, enabling it to start autonomous motor vehicles quicker and evolve to more recent, a lot quicker architectures.

Forging In advance

NIO is already off to the races with its program-defined lineup, saying programs to double the capacity of its plant in Hefei, China, to 240,000 vehicles for each year, with a facility capable of making up to 300,000.

As NIO scales creation capabilities and proceeds its expansion into world markets, NVIDIA HGX is scaling with them, enabling the deployment of one of the most innovative AI platforms in the automotive marketplace.

Leave a comment

Your email address will not be published.


*