A Precise Program Phase Identification Method Based on Frequency Domain Analysis

TL;DR

This paper introduces a method that converts program execution traces into the frequency domain to accurately identify program phases, aiding in optimization and prediction.

Contribution

It presents a novel frequency domain analysis technique for precise program phase identification and embedding phase info into code for better runtime prediction.

Findings

Accurately identifies program phases using frequency analysis.

Enables embedding phase info into program code.

Improves runtime phase prediction accuracy.

Abstract

In this paper, we present a systematic approach that transforms the program execution trace into the frequency domain and precisely identifies program phases. The analyzed results can be embedded into program code to mark the starting point and execution characteristics, such as CPI (Cycles per Instruction), of each phase. The so generated information can be applied to runtime program phase prediction. With the precise program phase information, more intelligent software and system optimization techniques can be further explored and developed.