Branch Prediction as a Reinforcement Learning Problem: Why, How and Case Studies

Anastasios Zouzias, Kleovoulos Kalaitzidis, Boris Grot

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Recent years have seen stagnating improvements to branch predictor (BP) efficacy and a dearth of fresh ideas in branch predictor design, calling for fresh thinking in this area. This paper argues that looking at BP from the viewpoint of Reinforcement Learning (RL) facilitates systematic reasoning about, and exploration of, BP designs. We describe how to apply the RL formulation to branch predictors, show that existing predictors can be succinctly expressed in this formulation, and study two RL-based variants of conventional BPs.
Original languageEnglish
Title of host publicationML for Computer Architecture and Systems 2021
Number of pages6
Publication statusAccepted/In press - 11 Jun 2021
EventML for Computer Architecture and Systems 2021 - Online
Duration: 19 Jun 202119 Jun 2021
https://sites.google.com/view/mlarchsys/isca-2021/

Workshop

WorkshopML for Computer Architecture and Systems 2021
Abbreviated titleMLArchSys 2021
Period19/06/2119/06/21
Internet address

Fingerprint

Dive into the research topics of 'Branch Prediction as a Reinforcement Learning Problem: Why, How and Case Studies'. Together they form a unique fingerprint.

Cite this