English

Evaluating Numerical Accuracy in Mixed-Precision Computing by Dual-Delta Testing

Numerical Analysis 2026-02-12 v1 Numerical Analysis Software Engineering

Abstract

Mixed-precision computing has become increasingly important in modern high-performance computing and machine learning applications. When implementing custom mixed-precision functions -- such as fused operators, optimized GPU kernels, or quantized inference paths -- it is critical to verify their numerical accuracy. Traditional approaches typically compare the custom implementation against a reference using a single error metric. However, this single-delta approach provides limited insight into whether the observed errors are inherent to the precision level or specific to the implementation. This paper introduces \textit{Dual-Delta Testing}, a systematic methodology that evaluates two error distributions against a high-precision oracle, enabling rigorous comparison between a custom implementation and a baseline reference. We present the mathematical framework, algorithmic formulation, statistical analysis techniques, and practical examples demonstrating the methodology's effectiveness in evaluating numerical accuracy.

Keywords

Cite

@article{arxiv.2602.10605,
  title  = {Evaluating Numerical Accuracy in Mixed-Precision Computing by Dual-Delta Testing},
  author = {Peichen Xie},
  journal= {arXiv preprint arXiv:2602.10605},
  year   = {2026}
}
R2 v1 2026-07-01T10:31:27.478Z