Tech giant NVIDIA’s GTC 2022 was held from March 21 to March 24. The conference brought together the best technology minds and discussed innovations across domains. NVIDIA CEO Jensen Huang made many major announcements in his keynote speech on March 22. As of now, the keynote video has crossed over 10 million views on YouTube.
In February, a month before the event, we did a story on what to expect from GTC 2022. Let us look at how our predictions played out.
Hopper architecture
NVIDIA officially announced the Hopper GPU microarchitecture and H100 GPU at GTC 2022 as the successor of NVIDIA Ampere architecture, launched two years back. NVIDIA H100 is packed with 80 billion transistors and has features such as Transformer Engine and a NVIDIA NVLink– ideal for working on large AI language models, recommender systems, and digital twins. The architecture is expected to rival the AMD Instinct MI 200 series, introduced last November.
Image: NVIDIA
NVIDIA H100 will be the engine of the world’s AI infrastructure that enterprises will use to accelerate their AI-driven businesses, said Jensen Huang.
H100 builds on the previous A100 model with improvements in architectural efficiency. As per the company, NVIDIA H100 can be deployed across data centre types such as on-premises, cloud, hybrid-cloud and edge. It will be available globally in the latter part of this year from cloud service providers as well as NVIDIA.
Industry has already gotten behind Hopper with leading names like Alibaba Cloud, Amazon Web Services, Baidu AI Cloud, Google Cloud, Microsoft Azure planning to offer H100-based instances.
DRIVE Hyperion 9
In CES 2022, held in January 2022, NVIDIA focused on DRIVE Hyperion 8 and highlighted the mass enterprise adoption of DRIVE Hyperion 8 platform. Autonomous vehicles were a major focus of GTC 2022.
In the keynote, Jensen announced the DRIVE Hyperion 9 open platform for automated and autonomous vehicles. It is built on multiple DRIVE Atlan computers to achieve intelligent driving and in-cabin functionality and will go into production in 2026. The DRIVE Hyperion 9 architecture comes with 14 cameras, nine radars, three lidars and 20 ultrasonics for automated and autonomous driving along with three cameras and one radar for interior occupant sensing.
Designed to be open and modular, the platform is compatible across generations, with the same computer form factor and NVIDIA DriveWorks APIs. The platform includes the computer architecture, sensor set and full NVIDIA DRIVE Chauffeur and Concierge applications.
Omniverse
NVIDIA is pushing the frontiers of metaverse. One of the major highlights of the event was the announcement of Omniverse Cloud, a suite of cloud services to provide artists, creators, designers and developers instant access to the NVIDIA Omniverse platform for 3D design collaboration and simulation from billions of devices.
NVIDIA said Nucleus Cloud is a one-click-to-collaborate sharing tool to help artists edit large 3D scenes from anywhere without transferring massive datasets. Omniverse Create is built for technical designers, artists and creators to build 3D worlds in real time. View app aids non-technical users to check Omniverse scenes streaming full simulation and rendering capabilities.
NVIDIA also announced computing system architecture, NVIDIA OVX, to run complex simulations in Omniverse. It will allow designers and engineers to create physically accurate digital twins and massive, true-to-reality simulation environments. Omniverse Code app will help with creating an integrated development environment for developers and help users to easily build their own Omniverse extensions, apps or microservices.