NVIDIA unveils new centralised computing technology for autonomous vehicles

NVIDIA has unveiled DRIVE Thor, a centralised car computer that is set to make the driving experience safer and easier for autonomous vehicles.

The new offering combines infotainment, automated driving and parking in a single, cost-saving system.

DRIVE Thor is said to achieve up to 2,000 teraflops of performance and unifies intelligent functions, including automated and assisted driving, parking, driver and occupant monitoring, digital instrument cluster, in-vehicle infotainment (IVI) and rear-seat entertainment.

It also includes the next-generation superchip and NVLink-C2C chip interconnect technology, which comes with advanced AI capabilities that were first introduced in the NVIDIA Hopper Multi-Instance GPU architecture.

The company is also putting a focus on efficiency and effectiveness. DRIVE Thor with MIG support for graphics and compute enables IVI and advanced driver-assistance systems to run domain isolation, allowing concurrent time-critical processes to run without interruption. 

"Advances in accelerated computing and AI are moving at lightspeed," says Jensen Huang, Founder and CEO, NVIDIA. 

"DRIVE Thor is the superhero of centralised compute, with lightning-fast performance to deliver continuously upgradable, safe and secure software-defined supercomputers on wheels."

Looking to the market, Geely-owned automaker ZEEKR has announced it will integrate DRIVE Thor on its centralised vehicle computer for next-generation intelligent electric vehicles, starting production in early 2025.

"ZEEKR users demand a luxury experience that includes the latest technology and safety features," says An Conghui, CEO, ZEEKR.

"NVIDIA DRIVE Thor will support our mission of providing cutting-edge technology that fulfills the needs of our customers and ensures ZEEKR remains at the forefront of tomorrow's innovations."

Typically with similar technology, dozens of electric control units are distributed throughout a vehicle to power individual functions. 

With DRIVE Thor, manufacturers can consolidate many functions on a single system-on-a-chip (SoC), which eases supply constraints and simplifies vehicle-design development, resulting in significantly lower cost, less weight and fewer cables.

Another advantage of DRIVE Thor is its 8-bit floating point (FP8) capability. Often developers lose neural-network accuracy when moving from 32-bit FP data to 8-bit integer format. NVIDIA says that DRIVE Thor features 2,000 teraflops of FP8 precision, allowing the transition to 8-bit without sacrificing accuracy. 

DRIVE Thor is also the first AV platform to incorporate an inference transformer engine, a new component of the Tensor Cores within NVIDIA GPUs. 

With this engine, DRIVE Thor can accelerate inference performance of transformer deep neural networks by up to 9x, which is paramount for supporting complicated AI workloads associated with self-driving.

​​The DRIVE Thor SoC and AGX board are developed to comply with ISO 26262 standards, and the software stack is designed for both ISO 26262 and ASPICE compliance.

"The shift to software-defined vehicles with centralised electronic architectures is accelerating, driving a need for more powerful and more energy-efficient compute platforms," says Sam Abuelsamid, principal research analyst, Guidehouse Insights. 

"The virtualisation, high-speed data transfer and massive processing performance of NVIDIA DRIVE Thor can enable safer vehicles, better user experiences and potential new revenue streams."

The tech will be available for automakers' 2025 models and will accelerate production roadmaps by bringing higher performance and advanced features to market in the same timeline.