Road to scalability for efficient graph search on massively parallel neuromorphic hardware

📅 2026-06-27
📈 Citations: 0
✨ Influential: 0
📄 PDF
🤖 AI Summary
This work addresses the challenge of efficiently and energy-efficiently computing shortest paths on large-scale weighted graphs using neuromorphic hardware. It proposes NEURO-MAPP, the first distributed algorithm that integrates the spiking neural network paradigm with shortest path computation. By leveraging the massive parallelism, local processing capabilities, and event-driven sparse communication mechanisms of the SpiNNaker 2 platform, NEURO-MAPP introduces a scalable graph search strategy tailored to neuromorphic architectures. Experimental results demonstrate that, across diverse synthetic and real-world graph datasets, NEURO-MAPP achieves substantially lower energy consumption compared to Dijkstra’s algorithm executed on conventional CPUs, while also exhibiting excellent runtime scalability—thereby overcoming performance bottlenecks that have historically limited traditional graph algorithms on neuromorphic hardware.
📝 Abstract
Efficient computation of shortest paths in weighted graphs is a fundamental problem with many applications. Neuromorphic hardware platforms promise massively parallel, efficient computation, changing parallelism tradeoffs. In this work, we introduce NEURO-MAPP (Neuromorphic-based Min-Add Parallel Propagation), a distributed shortest path algorithm designed to use the local computation and network communication available in neuromorphic systems. We provide an optimized implementation of the algorithm on the SpiNNaker 2 platform and evaluate its performance on a selection of synthetic and real-world graphs. These results are compared to Dijkstra's algorithm on a modern CPU. We find that the NEURO-MAPP implementation scales favorably in terms of runtime for many graph types while consuming less energy per shortest-path query than the CPU implementation in almost all cases. These findings highlight the potential of neuromorphic hardware featuring sparse, spike-based communication as a scalable and energy-efficient platform for computation in graph search and related tasks.
Problem

Research questions and friction points this paper is trying to address.

graph search
shortest path
neuromorphic hardware
scalability
massively parallel
Innovation

Methods, ideas, or system contributions that make the work stand out.

neuromorphic computing
graph search
shortest path algorithm
massively parallel
energy efficiency
🔎 Similar Papers
💼 Related Jobs
Authors to Follow
O
Oskar von Seeler
Department of Neuro- and Sensory Physiology, University Medical Center Göttingen, Germany; Campus Institute Data Science, University of Göttingen, Germany
E
Elena C. Offenberg
Third Institute of Physics – Biophysics, University of Göttingen, Germany
C
Carlo Michaelis
Third Institute of Physics – Biophysics, University of Göttingen, Germany
Tomas Kulvicius
Tomas Kulvicius
University of Goettingen, Third Institute of Physics - Biophysics, Dept. Computational Neuroscience
RoboticsMachine LearningArtificial Intelligence
J
Jannik Luboeinski
Department of Neuro- and Sensory Physiology, University Medical Center Göttingen, Germany; Campus Institute Data Science, University of Göttingen, Germany; Third Institute of Physics – Biophysics, University of Göttingen, Germany; Jülich Supercomputing Centre, Forschungszentrum Jülich, Germany
A
Andrew B. Lehr
Department of Neuro- and Sensory Physiology, University Medical Center Göttingen, Germany; Campus Institute Data Science, University of Göttingen, Germany; Third Institute of Physics – Biophysics, University of Göttingen, Germany; Circulant Labs, Bensheim, Germany
C
Christian Tetzlaff
Department of Neuro- and Sensory Physiology, University Medical Center Göttingen, Germany; Campus Institute Data Science, University of Göttingen, Germany; Third Institute of Physics – Biophysics, University of Göttingen, Germany