Learning Branch Probabilities in Compiler from Datacenter Workloads

TL;DR

This paper introduces a machine learning approach to estimate branch probabilities in compilers using datacenter workload data, outperforming heuristics and improving overall program performance.

Contribution

It presents a novel method for learning branch probabilities from large-scale data, enhancing compiler predictions without relying solely on profile-guided information.

Findings

Machine learning model reduces estimation error by 18-50%.

Achieves up to 8.1% performance gains on benchmarks.

Improves search application performance by over 1.2%.

Abstract

Estimating the probability with which a conditional branch instruction is taken is an important analysis that enables many optimizations in modern compilers. When using Profile Guided Optimizations (PGO), compilers are able to make a good estimation of the branch probabilities. In the absence of profile information, compilers resort to using heuristics for this purpose. In this work, we propose learning branch probabilities from a large corpus of data obtained from datacenter workloads. Using metrics including Root Mean Squared Error, Mean Absolute Error and cross-entropy, we show that the machine learning model improves branch probability estimation by 18-50% in comparison to compiler heuristics. This translates to performance improvement of up to 8.1% on 24 out of a suite of 40 benchmarks with a 1% geomean improvement on the suite. This also results in greater than 1.2% performance improvement in an important search application.