SpiNNTools: The Execution Engine for the SpiNNaker Platform
This provides a solution for researchers and engineers working with large-scale neuromorphic systems, though it appears incremental as it builds on existing SpiNNaker architecture.
The paper tackles the challenge of efficiently utilizing the SpiNNaker neuromorphic computing platform, which has millions of cores, by introducing SpiNNTools to map computational graphs into executable code, resulting in a toolset that addresses unique scalability issues in data loading, execution, and retrieval.
Distributed systems are becoming more common place, as computers typically contain multiple computation processors. The SpiNNaker architecture is such a distributed architecture, containing millions of cores connected with a unique communication network, making it one of the largest neuromorphic computing platforms in the world. Utilising these processors efficiently usually requires expert knowledge of the architecture to generate executable code. This work introduces a set of tools (SpiNNTools) that can map computational work described as a graph in to executable code that runs on this novel machine. The SpiNNaker architecture is highly scalable which in turn produces unique challenges in loading data, executing the mapped problem and the retrieval of data. In this paper we describe these challenges in detail and the solutions implemented.