Impact of GC design on power and performance for Android

Abstract

Small mobile devices have evolved to versatile computing systems. Android devices run a complete software stack, including a full Linux kernel, user land with several software daemons and a virtual machine to run applications. On these mobile systems energy is a scarce resource and needs to be handled carefully. Current systems rely on governors that adjust the frequency of individual cores depending on the system load. We measure energy consumption of different components of this complex software stack, including garbage collection (GC) of the Android virtual machine. Here we propose several extensions to the default GC configuration of Android, including a generational collector, across established dimensions of heap memory size and concurrency.Our evaluation shows that Android's asynchronous GC thread consumes a significant amount of energy. Therefore, varying the GC strategy can reduce total on-chip energy (by 20-30%) whilst slightly impacting application throughput (by 10-40%) and increasing worst-case pause times (by 20-30%). Our work quantifies the direct impact of GC on mobile system, enumerates the main factors and layers of this relationship, and offers a guide for analysis of memory behavior in understanding and tuning system performance.

SoK: The Challenges, Pitfalls, and Perils of Using Hardware Performance Counters for Security

“…First, we found mixed opinions on using HPCs: more than 45% of the papers in application areas that include performance analysis and optimizations, high performance computing and OS support do not recommend using HPCs. The main stated reasons for dismissing HPCs are the lack of determinism in performance counter values [10][11][12][13][14][15][16][17][18][19][20][21][22][23] and lack of portability of HPC events (e.g., some events may be present in one architecture but not in another [24][25][26][27]). These factors limit the applicability of performance counters in different application domains.…”

A Survey of Performance Optimization for Mobile Applications

“…However, a larger main memory size leads to higher energy consumption [40], [132]. Energy reductions can be achieved by using non-volatile memory [236], Phase Change Memory [40] or adaptations to the garbage collection [235] and scheduling algorithms [234].…”

Coastal Marine Data Crowdsourcing Using the Internet of Floating Things: Improving the Results of a Water Quality Model

“…In 2015, the architecture of the OSP project was completely redesigned in order to match the idea of a regular Android device that could act as a hub for IoFT-enabled sensors and actuators and released as FairWind in 2016, an Android application compliant with the Signal K standard, 6 which was being developed at the time as an open-source community standard for marine electronics data applications. Nevertheless, the behavior of the Android Garbage Collector management [26] and of the Android scheduler enforced by the Android Low Memory Killer [27], drastically affected the performance, and the real usability, of the FairWind application. That was due to FairWind's heavy CPU usage and RAM needs, which caused unrequited (and almost unpredictable) Android Service and background Activities process killing.…”