Implementing Deep Neural Networks (DNNs) on resource-constrained edge devices is a challenging task that requires tailored hardware accelerator architectures and a clear understanding of their performance characteristics when executing the intended AI workload. To facilitate this, we present an automated generation approach for fast performance models to accurately estimate the latency of a DNN mapped onto systematically modeled and concisely described accelerator architectures. Using our accelerator architecture description method, we modeled representative DNN accelerators such as Gemmini, UltraTrail, Plasticine-derived, and a parameterizable systolic array. Together with DNN mappings for those modeled architectures, we perform a combined DNN/hardware dependency graph analysis, which enables us, in the best case, to evaluate only 154 loop kernel iterations to estimate the performance for 4.19 billion instructions achieving a significant speedup. We outperform regression and analytical models in terms of mean absolute percentage error (MAPE) compared to simulation results, while being several magnitudes faster than an RTL simulation.
@article{arxiv.2409.08595,
title = {Automatic Generation of Fast and Accurate Performance Models for Deep Neural Network Accelerators},
author = {Konstantin Lübeck and Alexander Louis-Ferdinand Jung and Felix Wedlich and Mika Markus Müller and Federico Nicolás Peccia and Felix Thömmes and Jannik Steinmetz and Valentin Biermaier and Adrian Frischknecht and Paul Palomero Bernardo and Oliver Bringmann},
journal= {arXiv preprint arXiv:2409.08595},
year = {2025}
}
Comments
Accepted version for: ACM Transactions on Embedded Computing Systems