GTC Building digital twins is about to get a whole lot more affordable thanks to Nvidia, though some patience will be required.
Nvidia had a lot of news surrounding its Omniverse digital twinning platform to announce at its GPU Technology Conference (GTC) 2022. Nvidia VP/GM of enterprise visualization Bob Pette started with what amounts to an announcement of how well Omniverse has performed in the past year: “We believe our biggest growth driver this year will be Omniverse.”
The Omniverse ecosystem, launched last year, has grown by a factor of 10, said platform VP Richard Kerris, with major players like Unreal Engine, V-Ray, Blender, and others adding connections and a total of 26 CAD to USD file converters now available. Additional new connections for Omniverse, bringing the total to 82, include Bentley iTwin’s LumenRT and multiple new connections for Ipolog’s factory, logistics, and planning software.
The big new announcement: Omniverse Cloud
Of all the new features announced at GTC, Omniverse Cloud might be the most exciting. Nvidia said it’s exactly what it sounds like: “A suite of cloud services that gives artists, creators, designers, and developers instant access to the Nvidia Omniverse platform.”
Omniverse Cloud uses Nucleus (in this case, Nucleus Cloud) as the center of its software universe. Nvidia describes Nucleus Cloud as a “one-click-to-collaborate” tool that allows sharing and editing of large 3D scene files directly from the cloud. How do you edit and view those scenes? Via Omniverse Create and Omniverse View, the same apps used in on-prem Omniverse setups, but running in GeForce.
“What this really means is, whether you’re on a Chromebook, Mac, tablet or any other device that you use to connect to GeForce, now you’ll be able to stream, create, and view to your systems, and work and collaborate with other Omniverse customers,” Kerris said.
Unfortunately for those interested in jumping into the Omniverse Cloud, it’s still “under development.” There is a Nucleus Cloud early access program available, though.
Nvidia also announced new “industrial digital twin” hardware called OVX, which it said “is built to operate complex simulations that will run within Omniverse, enabling designers, engineers, and planners to create physically accurate digital twins and massive, true-to-reality simulation environments.”
OVX servers are big: eight A40 GPUs, three Nvidia ConnectX -6 Dx 200Gbps NICs, 1TB of system memory, and 16TB of NVMe storage. Those individual servers are configurable in pods of eight, all the way to an OVX SuperPOD made up of “32 OVX servers connected with Nvidia Spectrum-3 switch fabric or multiple OVX SuperPODs,” the company said.
OVX servers and SuperPODs will be available later this year from Inspur, Lenovo, and Supermicro.
Other Omniverse announcements
Omniverse Enterprise on Nvidia LaunchPad, a free turnkey app for enterprises looking to demo Omniverse capabilities, was expanded to nine regions, and additional labs and pre-populated scenes were added too.
A host of other new features and tools were added, and major updates include:
- Omniverse Code, a new IDE for building extensions, apps, and microservices for Omniverse
- DeepSearch, a new AI-based natural language search service for combing through untagged 3D asset libraries, now in early access for Omniverse enterprise customers
- Omniverse Replicator, a framework for accelerating perception network training
- Omniverse Avatar, announced in November 2021
- An Omniverse XR app is being developed for VR-optimized use of Omniverse View
- OmniGraph, ActionGraph, and AnimGraph are all getting major new releases
- Omniverse Kit, Create, View and Machinima are all getting new versions
There are a lot of little announcements, too, so be sure to read Nvidia’s full rundown of new Omniverse features. ®
See the original article here: The Register