NVIDIA announced DRIVE Thor, a centralized car computer to make the driving experience of self-driving cars safer and easier.
This new product combines infotainment, autonomous driving and parking into a single cost-saving system.
DRIVE Thor is said to deliver up to 2,000 teraflops of performance and intelligent features such as autonomous and assisted driving, parking, driver and occupant monitoring, digital instrument cluster, in-vehicle infotainment (IVI), and rear-seat entertainment. integrate the
It also includes next-generation superchips and NVLink-C2C chip interconnect technology with advanced AI capabilities first introduced in the NVIDIA Hopper multi-instance GPU architecture.
The company also focuses on efficiency and effectiveness. His DRIVE Thor with his MIG support for graphics and computing enables IVIs and advanced driver assistance systems to perform domain separation, allowing time-critical processes to run concurrently without interruption. Become.
NVIDIA founder and CEO Jensen Huang said:
“DRIVE Thor is a centralized computing superhero, delivering a lightning-fast performance, continuously upgradeable, safe and secure software-defined supercomputer on its wheels.”
Turning to the market, Geely-owned automaker ZEEKR has announced that it will integrate DRIVE Thor into its centralized vehicle computer for its next-generation intelligent electric vehicle, which will go into production in early 2025.
ZEEKR CEO An Conghui said:
“NVIDIA DRIVE Thor supports our mission to deliver cutting-edge technology that meets the needs of our customers, ensuring ZEEKR remains at the forefront of tomorrow’s innovation.”
Using similar technology, dozens of electrical control units are typically distributed throughout the vehicle to power individual functions.
DRIVE Thor allows manufacturers to integrate many functions into a single system-on-chip (SoC), easing supply constraints and simplifying the development of vehicle designs, resulting in significant cost savings, lightweight and reduce cables.
Another advantage of DRIVE Thor is its 8-bit floating point (FP8) capability. Developers often lose neural network precision when moving from 32-bit FP data to 8-bit integer format. According to NVIDIA, DRIVE Thor will feature 2,000 teraflops of his FP8 accuracy, allowing you to move to 8-bit without sacrificing accuracy.
DRIVE Thor is also the first AV platform to incorporate an inference transformer engine, a new component of Tensor Cores inside NVIDIA GPUs.
This engine enables DRIVE Thor to speed up the inference performance of Transformer deep neural networks by up to 9x. This is paramount to support complex AI workloads related to autonomous driving.
The DRIVE Thor SoC and AGX boards are developed to comply with the ISO 26262 standard and the software stack is designed to comply with both ISO 26262 and ASPICE.
Sam Abuelsamid, Principal Research Analyst at Guidehouse Insights, said:
“NVIDIA DRIVE Thor’s virtualization, high-speed data transfer, and massive processing power enable safer vehicles, better user experiences, and potential new revenue streams.”
The technology will be available for the automaker’s 2025 model year, accelerating production roadmaps by bringing higher performance and advanced features to market on the same timeline.
NVIDIA Unveils New Centralized Computing Technology for Self-Driving Cars
Source link NVIDIA Unveils New Centralized Computing Technology for Self-Driving Cars