A Controlled Diagnostic Study of Hardware-Induced Distortions in Hardware-Aware Training

TL;DR

This paper introduces a diagnostic framework to evaluate which hardware-induced distortions in AI accelerators can be mitigated through hardware-aware training, guiding hardware-software co-design.

Contribution

It models hardware non-idealities as structured perturbations and identifies diagnostics to determine their impact on training effectiveness.

Findings

Distinguishes perturbations that can be compensated by HAT from those that cannot.

Provides practical guidance for hardware-software co-design based on perturbation diagnostics.

Analyzes six types of hardware non-idealities and their effects on gradient-based optimization.

Abstract

Hardware-aware training (HAT) is widely used to improve the robustness of neural networks on non-ideal AI accelerators, such as analog in-memory computing (IMC) systems. However, not all hardware-induced distortions are equally compensable by training. This paper presents a diagnostic framework that models hardware non-idealities as structured perturbations of the forward operator and evaluates their compatibility with gradient-based optimization. We analyze six representative perturbation classes--read noise, variability, drift, stuck-at faults, IR-drop, and ADC discretization--and identify three key diagnostics: gradient expectation consistency, bounded gradient variance, and non-degenerate sensitivity. Our results show a clear separation between perturbations that can be compensated by HAT and those that consistently break optimization. This provides practical guidance for hardware-software co-design, clarifying which non-idealities can be addressed at the training level and which require circuit-, architecture-, or calibration-level mitigation. This study should be interpreted as a controlled empirical analysis under vanilla forward-perturbation HAT, rather than as a universal theory of hardware-aware training.