Industrial HPC has Arrived

High Performance Computing (or HPC) is the practice of aggregating multiple compute nodes to work in parallel to solve complex or large problems. HPC solutions have traditionally focused on scientific and engineering problems – for example weather modelling, genomic sequencing, fluid dynamic and aerodynamic modelling.

In recent years, use of GPUs in HPC clusters have increased processing power by many orders of magnitude. GPUs deliver massive numbers of cores, and replace thousands of CPUs in an HPC cluster, which allow larger datasets to be processed even faster while also being more energy efficient.

Along with the use of GPUs, Deep Learning models and software tuned for GPU processing, like NVIDIA’s CUDA stack, have enabled HPC clusters to tackle larger and more complex problems with even larger data sets.

Why HPC now?

In a word: Simulations. HPC solutions now enable realistic simulations to be developed in any industry, with any size dataset. This allows design and development iterations to be done in digital simulations, where problems can be identified and rectified, new designs realised and tested quickly. Bypassing the need for physical prototypes allows for faster iterations, higher quality results and less waste when items reach real world production. Simulations can be created for factories, production processes workflows, as well as individual components.

As Jensen Huang, NVIDIA CEO states in his keynote below, “The question for industry is no longer, what can computers do? The question is who will use it first to revolutionize their industry.”