Automatic estimation of performance requirements for software tasks of mobile devices

Schwarzer, Sarah; Peschlow, Patrick; Pustina, Lukas; Martini, Peter

doi:10.1145/1958746.1958796

“…While Schwarzer et al [44] suggest methods to estimate performance requirements of software tasks using simulation, our approach is based on the observation made by Sherwood et al [45] that a program's execution changes over time in phases. Our work characterizes the GC workload, which is common between all apps, as having a lower CPI compared to the average mutator workload.…”

Section: Related Workmentioning

confidence: 99%

Don't race the memory bus: taming the GC leadfoot

Hussein

¹

,

Hosking

²

,

Payer

³

et al. 2015

Proceedings of the 2015 International Symposium on Memory Management

View full text Add to dashboard Cite

Dynamic voltage and frequency scaling (DVFS) is ubiquitous on mobile devices as a mechanism for saving energy. Reducing the clock frequency of a processor allows a corresponding reduction in power consumption, as does turning off idle cores. Garbage collection is a canonical example of the sort of memory-bound workload that best responds to such scaling. Here, we explore the impact of frequency scaling for garbage collection in a real mobile device running Android's Dalvik virtual machine, which uses a concurrent collector. By controlling the frequency of the core on which the concurrent collector thread runs we can reduce power significantly. Running established multi-threaded benchmarks shows that total processor energy can be reduced up to 30 %, with end-to-end performance loss of at most 10 %.

show abstract

“…In this paper, we demonstrate that it is necessary to define GC requirements as a function of system mechanisms such as the governor and scheduling policies. While Schwarzer et al [44] suggest methods to estimate performance requirements of software tasks using simulation, our approach is based on the observation made by Sherwood et al [45] that a program's execution changes over time in phases. Our work characterizes the GC workload, which is common between all apps, as having a lower CPI compared to the average mutator workload.…”

Section: Related Workmentioning

confidence: 99%

Don't race the memory bus: taming the GC leadfoot

Hussein

¹

,

Hosking

²

,

Payer

³

et al. 2015

SIGPLAN Not.

2

0

View full text Add to dashboard Cite

Dynamic voltage and frequency scaling (DVFS) is ubiquitous on mobile devices as a mechanism for saving energy. Reducing the clock frequency of a processor allows a corresponding reduction in power consumption, as does turning off idle cores. Garbage collection is a canonical example of the sort of memory-bound workload that best responds to such scaling. Here, we explore the impact of frequency scaling for garbage collection in a real mobile device running Android's Dalvik virtual machine, which uses a concurrent collector. By controlling the frequency of the core on which the concurrent collector thread runs we can reduce power significantly. Running established multi-threaded benchmarks shows that total processor energy can be reduced up to 30 %, with end-to-end performance loss of at most 10 %.

show abstract