Introduction
SpiNNcloud Systems and the GWDG invite you to a Workshop on the SpiNNaker 2 hardware platform. Are you interested in novel hardware designs? Do you work in the fields of theoretical neuroscience, machine learning, robotics, or image and signal processing? Do you want to develop new software and find new ways to apply your research? Then don't miss this unique opportunity to get started with SpiNNaker!
The GWDG is glad to announce that it has acquired a SpiNNaker 2 system, which will be incorporated as a part of the KISSKI centre for artifical intelligence. SpiNNaker (which stands for Spiking Neural Network Architecture) belongs to the family of neuromorphic hardware. It tightly couples millions of ARM cores to efficiently simulate millions of individual spiking neurons, or large deep neural networks. As such it is an
ideal platform for brain simulations, as well as machine learning, signal processing and robotics.
Together with the developers of SpiNNaker, SpiNNcloud Systems, you will have the opportunity to be introduced to the platform, trial the system with a test board, and ask questions regarding the hardware and how you can combine it with your research.
The workshop will take place in the Rechenzentrum of the GWDG in Göttingen in a hybrid format, with a limited number of in-person seats. After the workshop we will look into providing further access to testing hardware, until the final installation of a SpiNNaker machine in Göttingen.
Figure 1: Example of a large SpiNNaker 2 configuration. See Paper 1 in References for source.
Key Points
- SpiNNaker hardware: Million ARM cores, fast, highly integrated, novel architecture
- Suited for simulation of neurons/brains, machine learning, robotics, signal processing, complex network behaviors
- Workshop: Hands-on learning directly with the developers
- Get on the ground floor of a new system at the GWDG!
Attendance
The Workshop will take place in a hybrid setup, in-person and also broadcast online. In-person attendance is preferred if you want the opportunity to interact with the test board, but places are also limited. Online attendants will be able to listen in on the presentations and make comments or questions. For in-person participants, please bring your own laptop, with ssh capabilities.
There are two registration forms, one for in-person and one for online attendance. Please only fill in one of them. Closer to the date of the Workshop we will contact participants in the online form if we still have places available for in-person attendance.
Reading Material and References
It is recommended to quickly leaf through the following references for the current state of the SpiNNaker hardware:
Paper 1: SpiNNaker 2: A 10 Million Core Processor System for Brain Simulation and Machine Learning
Paper 2: The SpiNNaker 2 Processing Element Architecture for Hybrid Digital Neuromorphic Computing
Short slide set: SpiNNaker 2: A Platform for Real-Time Bio-Inspired AI and Cognition
Presentation at TTI Vanguard, June 2023: Bio-inspiration for Cloude-level Hardware: Case Study SpiNNaker 2
Additionally, the following references for a previous version of the SpiNNaker hardware might also be helpful and are mostly still applicable:
Open Access book, "SpiNNaker: A Spiking Neural Network Architecture", 2020
Organizers
Martin Leandro Paleico (GWDG)
Alexander Goldmann (GWDG)
Dr. Christian Boehme (GWDG)
Matthias Lohrmann (SpiNNcloud Systems GmbH)
Mahmoud Aki (SpiNNcloud Systems GmbH)
Marco Stolba (SpiNNcloud Systems GmbH)
Prof. Dr. Christian Tetzlaff (University Medical Center Göttingen)
