Rules, Not Rankings: Interpretable AI for Processor Microarchitecture Design Analysis

IEEE Computer Architecture Letters, vol. 25, pp. 194–197

Abstract

Expert-driven microarchitecture analysis provides valuable insights but fails to scale to modern design spaces, while black-box AI offers scalability at the cost of interpretability. We present PTAH, a data-driven methodology that combines codified expert knowledge with an interpretable conditional-density model. PTAH derives human-readable rules that link processor mechanism configurations to the full IPC probability distribution. Clustering these rules yields design principles that highlight key interactions, trade-offs, and behavioral regimes. A case study on SPEC CPU workloads shows that PTAH uncovers design insights while preserving full interpretability.

Authors 3

1. Abdoulaye Gamatié

   Centre National de la Recherche Scientifique · Université de Montpellier · Laboratoire d'Informatique, de Robotique et de Microélectronique de Montpellier

   Affiliation as printed

   LIRMM, CNRS, University of Montpellier, Montpellier, France

2. Lincen Yang Aachen

   Leiden University

   Affiliation as printed

   LIACS, Leiden University, Leiden, The Netherlands

3. Paulo Pimenta

   Centre National de la Recherche Scientifique · Université de Montpellier · Laboratoire d'Informatique, de Robotique et de Microélectronique de Montpellier

   Affiliation as printed

   LIRMM, CNRS, University of Montpellier, Montpellier, France