Constructing next-generation clever automobiles requires an AI infrastructure that pushes the leading edge.
Electrical automobile maker NIO is utilizing NVIDIA HGX to construct a complete information middle infrastructure for creating AI-powered, software-defined automobiles. With high-performance compute, the automaker can repeatedly iterate on subtle deep studying fashions, creating sturdy autonomous driving algorithms in a closed-loop setting.
“The complicated situations confronted by mass-produced vehicles and the large quantity of knowledge these fleets generate are the cornerstones of NIO’s autonomous driving capabilities,” stated Bai Yuli, head of AI Platforms at NIO. “By utilizing NVIDIA high-performance compute options, NIO can speed up the trail to autonomous driving.”
NIO has already launched clever automobiles developed on this infrastructure, reminiscent of its absolutely electrical, clever flagship sedan, the ET7. Its mid-size efficiency sedan, the ET5 is scheduled to debut in September.
Along with high-performance information middle growth, each fashions are constructed on the Adam supercomputer, powered by 4 NVIDIA DRIVE Orin systems-on-a-chip. These automobiles characteristic autonomous driving and clever cockpit capabilities which can be repeatedly iterated upon and improved within the information middle for a revolutionary buyer expertise.
Constructing a Excessive-Efficiency AI Infrastructure With NVIDIA GPUs and Networking
The position of the info middle is to ingest, curate and label large quantities of knowledge for AI mannequin coaching at scale.
Information assortment fleets generate a whole bunch of petabytes of knowledge and billions of photographs every year. This information is then used to optimize the deep neural networks (DNNs) that may run within the automobiles.
NIO’s scalable AI infrastructure is powered by NVIDIA HGX with eight A100 Tensor Core GPUs and NVIDIA ConnectX-6 InfiniBand adapters. This scalable supercomputer cluster consists of NVME SSD servers and interconnects by means of the high-speed NVIDIA Quantum InfiniBand community platform.
This highly effective infrastructure permits massive quantities of deep studying coaching information to be transferred to supercomputer reminiscence or NVIDIA A100 video reminiscence at ultra-high speeds of as much as 200Gbps.
NVIDIA HGX A100 is a high-performance server platform designed for AI situations, together with massive datasets and complex fashions like those who energy autonomous automobiles. It incorporates a totally optimized NVIDIA AI software program stack in NGC.
The platform units a brand new compute density benchmark, condensing 5 petaflops of AI efficiency and replaces siloed infrastructures with a single platform for a variety of complicated AI purposes.
With the HGX A100, NIO is ready to flexibly develop and deploy scalable AI methods. It additionally permits the corporate to extend mannequin growth effectivity by as much as 20x, permitting it to launch autonomous automobiles sooner and evolve to newer, quicker architectures.
NIO is already off to the races with its software-defined lineup, saying plans to double the capability of its plant in Hefei, China, to 240,000 automobiles per yr, with a facility able to producing as much as 300,000.
As NIO scales manufacturing capabilities and continues its growth into international markets, NVIDIA HGX is scaling with them, enabling the deployment of one of the crucial superior AI platforms within the automotive trade.