What Is the Cost of Energy Monitoring? An Empirical Study on the Overhead of RAPL-Based Tools

TL;DR

This study quantifies the overhead caused by RAPL-based energy monitoring tools at high sampling rates and proposes strategies to reduce this overhead for more accurate energy profiling.

Contribution

It provides an empirical analysis of RAPL tool overhead and introduces optimized methods to minimize measurement impact at high sampling frequencies.

Findings

Existing tools can cause up to 46.75% time overhead at 1 kHz.

Our tools significantly reduce system call and math overhead.

Accessing RAPL values with lower-level instructions improves measurement speed.

Abstract

The Running Average Power Limit (RAPL) interface is widely used to estimate software energy consumption via CPU and DRAM counters, but tool design differences and high-frequency polling can introduce measurement overhead, namely, extra time and energy consumed by the tool itself.This paper quantifies the impact of RAPL-based tools on high-frequency (1 kHz) energy monitoring and investigates mitigation strategies. We conduct two controlled experiments: the first evaluates seven tools, including a user-space application and a kernel module developed by the authors, against a no-tool baseline, using six NAS Benchmark functions to quantify overhead. The second experiment isolates and times key functions for polling Model-Specific Registers (MSRs) (rdmsr and sys/proc_read) to estimate their execution latencies and identify potential slowdowns. The results show that existing user-space tools can introduce substantial time overhead at 1 kHz, whereas our tools significantly reduce system call overhead and inline math overhead. The time overhead of existing tools ranges from 0.25% to 46.75%. Our solutions maintain time overhead levels close to the baseline. We also find that system calls are slower than rdmsr, which in turn is slower than traditionally long-running instructions like cpuid. These findings indicate that RAPL-based energy measurement can be substantially improved by simplifying tool design and employing lower-level instructions to access RAPL values. Our findings provide guidance for practitioners on how to develop high-frequency energy profiling tools, show possible situations that can skew energy values, and demonstrate that access to RAPL values can be faster using specific techniques.