Making Multi-GPU Simulation Accessible to All with NVIDIA DGX A100

Summary

Barracuda Virtual Reactor is the industry-standard tool for simulating fluidized bed reactors (FBRs) for many applications in refining, petrochemicals, cement manufacture, power generation, and other energy-intensive industries. The complexity of FBR simulation lends itself well to acceleration via GPU computing. Virtual Reactor utilizes NVIDIA’s multi-GPU computing technology to achieve 50-400x speedups over serial processing. This session demonstrates how the NVIDIA DGX A100 provides multi-GPU compute capabilities to everyday software users, in a form factor that fits under a desk or compactly in the data center. Week-long simulations can now be completed in an hour, and previously intractable problems are now solved in industrially relevant time scales.

About the Speaker

Andy Lin is VP of Strategy & Innovation at Mark III Systems, an NVIDIA NPN Elite and CPFD partner, with a strong focus and unique team partnering around accelerated computing strategies for the enterprise, including AI/ML, HPC, simulation, visualization, and virtualization workloads. Mark III’s unique “full stack” team comprises of systems engineers, data scientists, devs, and DevOps/MLOps engineers that work together to ensure the success of organizations, including those deploying and operating Barracuda Virtual Reactor on NVIDIA platforms.