NVIDIA at SC21

At 100,000 times faster than traditional numerical weather models, the Fourier Neural Operator (FNO) -- a novel AI model powered by NVIDIA GPUs,  is a significant step towards building a digital twin of the Earth.

With NVIDIA Omniverse’s rendering, simulation, and GPU-acceleration technologies, climate researchers at the University of Wisconsin are moving one step closer to understanding the complexity of these unpredictable storms with collaborative, interactive scientific workflows.

Using NVIDIA Modulus physics-ML framework and NVIDIA Omniverse, we’ve trained high fidelity, physics-based, neural network models, and created a sophisticated digital twin platform with Siemens Energy.

Traditional methods for performing data reductions are very costly in terms of latency and CPU cycles. The NVIDIA Quantum InfiniBand switch with SHARP technology addresses complex operations. NVIDIA Quantum switches perform this data reduction in a fraction of the time as traditional methods.

Magnum IO GPUDirect over an InfiniBand network enables Verizon’s breakthrough distributed volumetric video architecture. By placing their technology into edge computing centers located at sports centers around the United States and in Verizon facilities, they’re able to bring 3D experiences to media and serve up new options for putting you in the game.

Creating a “Digital Twin” of the Earth using technologies like the NVIDIA Omniverse ParaView Connector allows scientists to interactively analyze weather and climate data in 3D, and to combine information from multiple sources and explore “what if” scenarios.