Spatial accelerators, composed of arrays of compute-memory integrated units, offer an attractive platform for deploying inference workloads with low latency and low energy consumption. However, fully exploiting their architectural advantages typically requires careful, expert-driven mapping of computational graphs to distributed processing elements. In this work, we automate this process by framing the mapping challenge as a black-box optimization problem. We introduce the first evolutionary, hardware-in-the-loop mapping framework for neuromorphic accelerators, enabling users without deep hardware knowledge to deploy workloads more efficiently. We evaluate our approach on Intel Loihi 2, a representative spatial accelerator featuring 152 cores per chip in a 2D mesh. Our method achieves up to 35% reduction in total latency compared to default heuristics on two sparse multi-layer perceptron networks. Furthermore, we demonstrate the scalability of our approach to multi-chip systems and observe an up to 40% improvement in energy efficiency, without explicitly optimizing for it.
@article{arxiv.2602.04717,
title = {Evolutionary Mapping of Neural Networks to Spatial Accelerators},
author = {Alessandro Pierro and Jonathan Timcheck and Jason Yik and Marius Lindauer and Eyke Hüllermeier and Marcel Wever},
journal= {arXiv preprint arXiv:2602.04717},
year = {2026}
}