gAIn meets industry innovation: SpiNNcloud workshop hosted at LRZ

Proud to share highlights from the recent SpiNNcloud workshop hosted at LRZ

The event brought together LRZ (Leibniz-Rechenzentrum), HLRS (Höchstleistungsrechenzentrum Stuttgart), and SpiNNcloud to exchange knowledge on infrastructure, neuromorphic hardware, and future research directions.

A key focus of our gAIn – Next Generation AI Computing project is exploring innovative, energy-efficient hardware, such as the brain-inspired SpiNNaker2 processor. gAIn researchers joined this workshop with the Dresden-based startup SpiNNcloud to dive into neuromorphic computing architectures.

Neuromorphic computing models the structure and dynamics of biological neural networks, enabling high-throughput processing with significantly reduced energy consumption—particularly effective for sparse and event-driven workloads. The team explored how neuromorphic chips like SpiNNaker2 can be applied to high-impact use cases in numerical analysis and combinatorial optimization, achieving state-of-the-art accuracy with substantial energy savings compared to conventional GPU-based methods.

Workshop agenda:
🔹 Introduction to LRZ and BEAST (Bavarian Energy, Architecture, and Software Testbed)
🔹 Introduction to HLRS
🔹 Deep dive into SpiNNcloud – infrastructure, neuromorphic hardware, and the SpiNNaker concept: sparse processing, energy-proportional computing, robust prediction, highly parallel and asynchronous processing
🔹 Research on SpiNNaker 2 and Hybrid Systems

In the opening, LRZ and HLRS presented their work, areas for improvement, and potential ways SpiNNcloud hardware could be integrated into their systems. SpiNNcloud followed with an overview of their applications, publications, and expansion plans, including a sneak peek at their API stack.

The workshop concluded with an open discussion where participants explored collaboration opportunities, asked in-depth technical questions, and received valuable insights from SpiNNcloud’s team.

We look forward to continuing our collaboration and working on ways to scale applications in sequence modeling and generative AI, further pushing forward the boundaries of neuromorphic computing.