AI Insight
This paper introduces a novel method for constructing large-scale spiking neural networks (SNNs) on multi-GPU clusters and exascale supercomputers, using the Message Passing Interface (MPI) to distribute network building and spike communication across thousands of GPUs. Inspired by the structural properties of the human cerebral cortex — a sparse network of roughly 10 billion neurons each connected via thousands of synapses — the approach allows each MPI process to independently build its local connectivity and prepare efficient data structures for inter-GPU spike exchange. The authors validate their method by demonstrating scaling performance on two established cortical models, using both point-to-point and collective communication strategies.
Why it matters
Efficiently simulating biologically realistic neural networks at brain-scale is a fundamental challenge in computational neuroscience; this work directly enables researchers to run larger and more detailed brain models on next-generation supercomputing infrastructure, potentially accelerating discoveries in areas such as neural circuit dynamics, brain-inspired AI, and neurological disorder modeling.
arXiv:2512.09502v2 Announce Type: replace-cross
Abstract: Diverse scientific and engineering research areas deal with discrete, time-stamped changes in large systems of interacting delay differential equations. Simulating such complex systems at scale on high-performance computing clusters demands efficient management of communication and memory. Inspired by the human cerebral cortex — a sparsely connected network of $mathcal{O}(10^{10})$ neurons, each forming $mathcal{O}(10^{3})$–$mathcal{O}(10^{4})$ synapses and communicating via short electrical pulses called spikes — we study the simulation of large-scale spiking neural networks for computational neuroscience research. This work presents a novel network construction method for multi-GPU clusters and upcoming exascale supercomputers using the Message Passing Interface (MPI), where each process builds its local connectivity and prepares the data structures for efficient spike exchange across the cluster during state propagation. We demonstrate scaling performance of two cortical models using point-to-point and collective communication, respectively.
Source: Scalable Construction of Spiking Neural Networks using up to thousands of GPUs