NVIDIA’s GPU Technology Conference (GTC) event concluded on April 16. During the conference, several announcements were made, including the launch of data centre chip Grace, Omniverse platform, NVIDIA EGX enterprise platform, and cuQuantum.
The founder and CEO Jensen Huang also unveiled NVIDIA DRIVE Atlan. Atlan centralises the vehicle’s computing infrastructure into a single system-on-a-chip (SoC) and offers AI compute for autonomous vehicles. Earlier, NVIDIA had released a centralised computing platform, DRIVE, for autonomous vehicles that fuse AI and software with state-of-the-art networking and security solutions.
Atlan: Data centre on wheels
The global chip shortage has badly hit several industries, especially the automobile industry.
While announcing Atlan, CEO Huang said, “Covid exposed a weakness in the supply chain of the automotive industry, which has too many components that go into cars.” Huang said the challenge could be handled by having centralised components. Future vehicles will need an optimised AI architecture for autonomous driving, speech recognition and driver monitoring.
The newly introduced Atlan, slated for public release in 2025, achieves similar centralisation. As per NVIDIA, Atlan can deliver more than 1,000 TOPs. Atlan leverages NVIDIA’s latest GPU architecture, new Arm CPU cores, deep learning and computer vision algorithms to deliver four times better performance compared to the previous generation. Further, the platform architecture offers great compute horsepower for future AI vehicles, and allow developers to add features and other updates. Atlan will be able to concurrently run autonomous vehicle, intelligent cockpit, and infotainment applications.
Atlan will have the NVIDIA BlueField data processing unit (DPU). BlueField DPUs provide a secure and accelerated infrastructure for a workload in the cloud, data centre or edge environments. It combines powerful computing, full infrastructure-on-chip programmability, and high-performance networking. The integration with BlueField DPU will allow Atlan to run a large number of AI applications seamlessly. BlueField DPU will offer a zero-security architecture with protection against cyber-attacks, intrusion detection, and other critical security features.
Atlan is programmable through CUDA and TensorRT APIs and libraries.
About NVIDIA DRIVE
“The transportation industry needs a computing platform that it can rely on for decades. NVIDIA DRIVE is the most advanced AI and AV computing platform, with rich global software and developer ecosystems, and architecturally compatible for generations,” said Jensen Huang.
NVIDIA DRIVE is an autonomous vehicle platform introduced at the Consumer Electronics Show (CES) in 2015. It offers a full-stack solution
for highly automated, supervised driving. It includes safety features, automated driving, and parking and AI cockpit capabilities ranging from Level 2 to Level 5. Components of NVIDIA Drive include software platform AP2X, Xavier and Orin (SoC hardware), Driver Interface (IX), and Datacentre Deep Learning (DGX).
In 2017, Intel acquired Israeli company Mobileye that develops self-driving vehicles and advanced driving assistance systems. At CES 2018, Mobileye launched fifth-generation SoC EyeQ5 for fully autonomous
driving. It achieves 24 TOPS at 10W and operates level 4 and 5 autonomy. EyeQ5 can perform sensor fusion to process data from various sensors. It competes with NVIDIA’s Xavier SoC.
Recently, Intel-Mobileye announced their latest offering — LiDAR on-chip. It offers Level 4 autonomy and operates a lidar vision system and a separate camera subsystem.
Qualcomm launched its autonomous driving solution Snapdragon Ride, at CES 2020. Snapdragon Ride leverages its power-efficient hardware, AI technologies, and autonomous driving stack to deliver a cost and energy-efficient system. Before this, Qualcomm’s offerings in the autonomous driving space were limited to supplying infotainment solutions.
Tesla introduced a custom chip called the full self-driving computer during its Autonomy Day in 2019. Built in collaboration with Samsung, Tesla said FSD is a high-performance and special-purpose chip that would solve several challenges faced by self-driven vehicles.
Join Our Telegram Group. Be part of an engaging online community. Join Here.
Subscribe to our NewsletterGet the latest updates and relevant offers by sharing your email.
I am a journalist with a postgraduate degree in computer network engineering. When not reading or writing, one can find me doodling away to my heart’s content.