Sensitivity Analysis of Core Specialization Techniques

TL;DR

This paper evaluates five core specialization techniques to improve performance of OS-intensive workloads with instruction footprints exceeding cache size, finding SchedTask generally performs best across various configurations.

Contribution

It provides a comparative analysis of five state-of-the-art core specialization methods for OS-intensive workloads, highlighting SchedTask's superior performance.

Findings

SchedTask outperforms other techniques across tested workloads.

Core specialization significantly reduces instruction cache misses.

Performance varies with system parameters and workload characteristics.

Abstract

The instruction footprint of OS-intensive workloads such as web servers, database servers, and file servers typically exceeds the size of the instruction cache (32 KB). Consequently, such workloads incur a lot of i-cache misses, which reduces their performance drastically. Several papers have proposed to improve the performance of such workloads using core specialization. In this scheme, tasks with different instruction footprints are executed on different cores. In this report, we study the performance of five state of the art core specialization techniques: SelectiveOffload [6], FlexSC [8], DisAggregateOS [5], SLICC [2], and SchedTask [3] for different system parameters. Our studies show that for a suite of 8 popular OS-intensive workloads, SchedTask performs best for all evaluated configurations.