Sparxity Announces Multi-Million-Dollar Sparse-Native Computing Deployment at the Jülich Supercomputing Centre
aiinsider ·
Sparxity, formerly known as SpiNNcloud, has deployed a multi-million-dollar sparse-native computing system at the Jülich Supercomputing Centre in Germany. The new event-based infrastructure, named JUBIS, is designed to compute only when activity occurs, thereby reducing computational workloads, data movement, and energy consumption. This deployment aims to execute sparse solvers ten times faster while using eighteen times less energy for event-based machine learning tasks. Additionally, the system promises to reach acceptable optimization solutions twenty times faster than traditional dense hardware methods. Dr. Guido Trensch from the Jülich Supercomputing Centre highlighted that this approach opens new research opportunities in neuroscience by complementing existing high-performance computing capabilities. Sparxity CEO Hector Gonzalez emphasized that the company’s rebrand focuses on unlocking AI potential by removing the constraints of dense hardware limits. The installation serves as a proof point for sparse-native computing in real-world supercomputer environments and is supported by European regional funding programs.
aiinsider