An End-to-End DNN Inference Framework for the SpiNNaker2 Neuromorphic MPSoC

2025-07-18

AI Summary

This work addresses the challenge of deploying large, complex deep neural networks (DNNs)—notably Transformers—on the SpiNNaker2 neuromorphic MPSoC for efficient inference. We present the first end-to-end PyTorch-to-SpiNNaker2 inference framework, built upon an extended OctopusScheduler that integrates multi-layer DNN scheduling, fine-grained model quantization, operator downgrading, and neuromorphic hardware mapping. Our framework enables full compilation and execution of PyTorch models directly onto a single SpiNNaker2 chip. To our knowledge, this is the first demonstration of end-to-end Transformer-scale DNN inference on SpiNNaker2, achieving high accuracy while significantly improving energy efficiency and throughput. The work overcomes a key bottleneck in neuromorphic hardware support for large-scale DNN inference and provides a scalable, system-level solution for deploying complex AI models at the edge.

Technology Category

• Machine Learning: Hardware-aware ML
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

This work presents a multi-layer DNN scheduling framework as an extension of OctopuScheduler, providing an end-to-end flow from PyTorch models to inference on a single SpiNNaker2 chip. Together with a front-end comprised of quantization and lowering steps, the proposed framework enables the edge-based execution of large and complex DNNs up to transformer scale using the neuromorphic platform SpiNNaker2.

Problem

• Develop end-to-end DNN inference for SpiNNaker2 neuromorphic chip
• Enable edge-based execution of large-scale DNNs
• Extend OctopuScheduler with PyTorch-to-SpiNNaker2 flow

Innovation

• End-to-end DNN scheduling framework for SpiNNaker2
• PyTorch to neuromorphic chip inference flow
• Supports transformer-scale DNNs on edge