Cadence brings AI supercomputing to digital twin simulations

Cadence Design Systems has launched a hardware and software platform to bring supercomputing performance to digital twin and AI simulations.

San Jose, California-based Cadence has focused on tools for electronic systems design. And today it is announcing its Cadence Millennium Enterprise Multiphysics Platform, which it says is a big leap forward in simulation.

The platform leverages GPU and CPU computing to deliver up to 100 times the design impact through accuracy, speed and scale.

GPU-resident solvers dramatically improve the energy efficiency of simulation by 20 times over
traditional high-performance computing. Cadence said the digital twin, AI and HPC technology combine to provide the best multiphysics simulation solution for automotive, A&D, energy, turbomachinery and data center markets.

“Throughout our 35-year history, Cadence has been focused on increasing performance with no loss of accuracy for the most challenging computational applications. Algorithmic throughput remains a key priority, and we’re now leveraging generative AI to harvest knowledge gained from massive quantities of design and simulation data,” said Ben Gu, corporate vice president of R&D for multiphysics systems analysis at Cadence. “Our revolutionary Millennium platform is a giant leap forward, delivering unprecedented acceleration and scalability of digital twin and AI applications. CFD is poised to benefit greatly from performance and efficiency gains, and the tremendous power of the Millennium M1 is disrupting industries that must explore more design innovations and bring them to market faster.”

The hardware/software combo will make it easier to create digital twins, or physically accurate virtual imitations of real-life factories and other physical creations.

The platform integrates graphics processing units (GPU) and central processing unit (CPU) computing with proprietary software algorithms, ushering in better performance and energy efficiency for multiphysics system design and analysis.

The first-generation Cadence Millennium M1 focuses on accelerating high-fidelity computational fluid dynamics (CFD) simulations, catering to industries such as automotive, aerospace and defense (A&D), energy, and turbomachinery. The platform aims to address the increasing demand for designing mechanical systems that achieve optimal performance and efficiency, aligning with industry priorities such as improving fuel efficiency, reducing emissions, and enhancing overall system functionality.

Features and benefits

Performance Revolution: Combines GPU-resident CFD solvers with dedicated GPU hardware, achieving supercomputer-equivalent throughput per GPU of up to 1, 000 CPU cores. It ensures optimal performance for complex mechanical systems with near-linear scalability on multiple GPU nodes.

Efficiency overhaul: Reduces simulation turnaround time from weeks to hours, boasting 20 times better energy efficiency compared to traditional CPU equivalents. Drives energy-efficient simulation, a critical aspect for industries prioritizing sustainability.

Accuracy and fidelity: Leverages Cadence Fidelity CFD solvers, providing unmatched accuracy to tackle complex simulation challenges. Ensures high-fidelity simulation data crucial for building reliable digital twin visualizations.

AI digital twin integration: Employs generative AI technology for rapid generation of high-quality multiphysics data. Enables the creation of fast and reliable digital twin visualizations for optimal system design solutions.

Turnkey solution: Couples GPU compute with modern and scalable CFD solvers, offering an optimized environment for accelerated CFD and multidisciplinary design and optimization. Provides flexibility with deployment options, available in the cloud or on premises, accommodating various customer needs.

The company received endorsements from big companies and academics, including Honda, Corvid Technologies, Asus and Stanford University’s Parviz Moin.

“Honda and Cadence have worked together for more than 10 years to accelerate CFD design
processes,” said Atsushi Ogawa, COO at Honda R&D Co., in a statement. “Our deployment of Cadence’s Millennium M1 multiphysics supercomputing platform, with its LES solver, has demonstrated significantly faster turnaround time and the scalability to simulate much larger systems for aerodynamics, combustion and aeroacoustics applications for our automotive, aeropropulsion and eVTOL projects. With the potential for significant acceleration of high-fidelity CFD simulations, we look forward to continuing our collaboration to develop next-generation CFD solutions.”