The collaboration aims to bring AI-driven simulation capabilities to engineers and designers.
SAN FRANCISCO—Rescale, a cloud high performance computing (HPC) platform for AI-accelerated modeling and simulation, reported in November that it is collaborating with NVIDIA to bring AI-driven, real-time simulation capabilities to engineers and designers.
The collaboration involves integrating Rescale’s cloud-native platform with the NVIDIA Omniverse Blueprint for real-time computer-aided engineering digital twins. This integration “helps set a new industry standard for AI-based CAE acceleration alongside improvement in efficiency, productivity, and speed in design workflows by enabling the rapid creation and deployment of AI-powered digital twins,” according to a release from Rescale.
“Our collaboration with Rescale brings scalable cloud resources to help unlock the potential of real-time simulation in the engineering and design space,” said Tim Costa, senior director of CAE, electronic design automation, and quantum at NVIDIA, in the release. “This ability to achieve near-instantaneous feedback on design changes makes complex workflows more intuitive and accessible—marking a milestone in AI-powered engineering.”
The integration combines Rescale AI’s advanced capabilities with NVIDIA’s accelerated computing technologies, as well as NVIDIA Modulus for pre-trained physics-based surrogate models; NVIDIA-powered application programming interfaces for connecting complex computer-aided engineering (CAE) workflows; NVIDIA NIM microservices for accelerated deployment and real-time inferencing; and NVIDIA Omniverse APIs for interactive visualization, the release stated.
Novid Beheshti, engineering specialist CAE at Intelligent Energy, said the AI-based, computer-aided engineering (CAE) approach is “a game-changer” for optimizing Intelligent Energy’s modeling and simulation workflows and a “key enabler for improving the efficiency” of its R&D efforts.
“The NVIDIA Omniverse Blueprint on Rescale enables Intelligent Energy to tackle complex challenges in hydrogen fuel cell development,” Beheshti said in the release. “By reducing complex simulation times from 24 hours to mere milliseconds and preprocessing times from 8 hours down to milliseconds, Rescale’s platform transforms our design process, enabling engineers to receive real-time feedback and significantly accelerating the delivery of next-generation solutions.”
By leveraging Rescale’s open-architecture high performance computing cloud platform, engineers reportedly can run “any CAE application on any cloud provider to generate synthetic design data and train AI models, creating AI-powered digital twins that provide real-time feedback and reduce design iteration cycles,” according to the release.
“Training AI models on Rescale is simple, turning every simulation engineer into a data scientist. First, engineers can run simulations used as training data, which is automatically processed and structured for model training. Next, they can use Rescale’s built-in AI training framework powered by NVIDIA Modulus. Finally, models can be trained and published directly to the Rescale AI Model Library, where they are readily accessible for immediate use in simulations across teams,” the release stated.
NVIDIA Omniverse, a development platform for real-time data interoperability and physically based visualization, is embedded into Rescale AI, minimizing data movement during model creation and allowing engineers to visualize their AI simulations instantly, according to Rescale. This is said to enable the rapid creation of innovations like an AI-powered wind tunnel, as the model can be trained in a week instead of the typical months of back-and-forth between data science teams, making instant feedback loops and continuous design improvements possible, the company said in the release.
“This collaboration represents a major leap forward in CAE, pushing the boundaries of what’s possible in real-time simulation,” said Joris Poort, CEO of Rescale, in the release. “Through leveraging NVIDIA’s advanced GPU technologies, CUDA-X libraries, Modulus, the NVIDIA Warp developer framework, NIM microservices, and Omniverse, combined with Rescale’s high performance compute cloud platform, industry-leading R&D application catalog, and suite of Rescale Data and Rescale AI capabilities, we’re enabling engineers and designers to create, test, and iterate designs faster than ever before.
“This real-time feedback loop empowers teams to make smarter decisions and drive innovation in product development with unprecedented efficiency,” Poort continued. “Looking ahead, combining AI simulation with AI-enabled agents will provide real-time multi-disciplinary product development feedback, creating a unified engineering ecosystem where human ingenuity and AI come together to achieve highly optimal designs.”