Rescale AI Models Use NVIDIA Omniverse Blueprint

The Omniverse Blueprint from NVIDIA

The Omniverse Real-Time Physics Digital Twins have been announced by NVIDIA in partnership with leading software companies. For Altair, Ansys, Cadence, Siemens, and other businesses, this design for interactive virtual wind tunnels enables computer-aided engineering exploration that has never been done before.

SC24- Industry software developers can help their computer-aided engineering (CAE) clients in the manufacturing, energy, automotive, aerospace, and other sectors create dynamic, real-time digital twins with the NVIDIA Omniverse Blueprint, which was revealed today.

Software developers like as Altair, Ansys, Cadence, and Siemens may help their clients save development costs and energy usage while speeding time to market by utilizing the NVIDIA Omniverse Blueprint for real-time computer-aided engineering digital twins. The blueprint is a standard technique that uses interactive physically based rendering, NVIDIA acceleration libraries, and physics-AI frameworks to achieve 1,200x faster simulations and real-time visualization.

The purpose of Omniverse was to make it possible to construct digital twins of anything. Among the first applications of the blueprint are computational fluid dynamics (CFD) simulations, an essential first step in digitally investigating, testing, and refining the designs of cars, airplanes, ships, and many other products. Completing standard engineering tasks, such as physics simulation, visualization, and design optimization, might take weeks or even months.

At SC24, NVIDIA and Luminary Cloud are demonstrating an industry-first virtual wind tunnel that allows users to simulate and observe fluid dynamics at real-time, interactive speeds, even while changing the vehicle model within the tunnel.

Unifying NVIDIA's Three Technology Foundations for Developers

Building a real-time physics digital twin requires two key competencies: real-time visualization of large-scale data and real-time physics solver performance.

To achieve these, the Omniverse Blueprint integrates the NVIDIA CUDA-X libraries to accelerate the solvers, the NVIDIA Modulus physics-AI framework to train and implement models to generate flow fields, and the NVIDIA Omniverse application programming interfaces for 3D data interoperability and real-time RTX-enabled visualization.

The developers' existing tools can either fully or partially include the plan.

Ecosystem Advances Simulations using NVIDIA Blueprint

Ansys was the first to incorporate the NVIDIA Omniverse Blueprint in their fluid simulation application, Ansys Fluent, to enable fast CFD simulation.

Ansys ran Fluent on 320 NVIDIA GH200 Grace Hopper Superchips at the Texas Advanced Computing Center. A 2.5-billion-cell automobile simulation that would have taken more than a month to run on 2,048 x86 CPU cores was completed in little over six hours. This established a new industry standard and significantly improved the feasibility of overnight high-fidelity CFD testing.

The combination of NVIDIA Omniverse Blueprint with Ansys software allows the clients to process more complex and sophisticated simulations more quickly and accurately. "The partnership is improving design and engineering standards across multiple industries."

Luminary Cloud is also using the model. The company's new simulation AI model, which is based on NVIDIA Modulus, understood the relationships between airflow fields and vehicle geometry using training data from their GPU accelerated CFD solver. The model performs simulations orders of magnitude faster than the solver itself, enabling real-time aerodynamic flow simulation using Omniverse APIs.

Other companies looking at integrating the Omniverse Blueprint into their own systems include Siemens, SimScale, Altair, Beyond Math, Cadence, Hexagon, Neural Concept, and Trane Technologies.

The Omniverse Blueprint is interoperable with all of the leading cloud computing platforms, including Oracle Cloud Infrastructure, Google Cloud, Microsoft Azure, and Amazon Web Services. It is also provided via NVIDIA DGX Cloud.

Rescale, a cloud-based platform that helps companies accelerate scientific and engineering breakthroughs, is using the NVIDIA Omniverse Blueprint to enable companies to train and deploy customized AI models with a few clicks.

The Rescale platform automates the whole application-to-hardware stack and may be used with any cloud service provider. Any simulation solver may be used by businesses to build, train, and deploy AI models, generate training data, execute inference predictions, and show and optimize models.