Performance and power profiling for emulated Android systems

Abstract

Simulation is a common approach for assisting system design and optimization. For system-wide optimization, energy and computational resources are often the two most critical issues. Monitoring the energy state of each hardware component and measuring the time spent in each state is needed for accurate energy and performance prediction. For software optimization, it is important to profile the energy and the time consumed by each software construct in a realistic operating environment with a proper workload. However, the conventional approaches of simulation often fail to produce satisfying data. First, building a cycle-accurate simulation environment for a complex system, such as an Android smartphone, is difficult and can take a long time. Second, a slow simulation can significantly alter the behavior of multithreaded, I/O-intensive applications and can affect the accuracy of profiles. Third, existing software-based profilers generally do not work on simulators, which makes it difficult for performance analysis of complicated software, for example, Java applications executed by the Dalvik VM in an Android system. To address these aforementioned problems, we proposed and prototyped a framework, called virtual performance analyzer (VPA). VPA takes advantage of an existing emulator or virtual machine monitor to reduce the complexity of building a simulator. VPA allows the user to selectively and incrementally integrate timing models and power models into the emulator with our carefully designed performance/power monitors, tracing facility, and profiling tools to evaluate and analyze the emulated system. The emulated system can perform at different levels of speed to help verify if the profile data are impacted by the emulation speed. Finally, VPA supports existing software-based profiles and enables non-intrusive tracing/profiling by minimizing the probe effect. Our experimental results show that the VPA framework allows users to quickly establish a performance/power evaluation environment and gather useful information to support system design and software optimization for Android smartphones.

Alea

“…Other energy profiling tools build instruction-level power models bottom-up from gate-level or design-time models to provide power profiles to simulators and hardware prototyping environments [14,13]. These inherently static models fail to capture the variability in instruction-level power consumption due to the context in which instructions execute.…”

“…Prior energy accounting tools can be broadly classified into two categories: tools that directly measure energy using on-board sensors or external instruments [1,2,3,4,5,6]; and tools that model energy based on activity vectors derived from hardware performance counters, kernel event counters, finite state machines, or instruction counters in microbenchmarks [7,8,9,10,11,12,13,14,15,16,17]. All of these tools can associate energy measurements with software contexts via manual instrumentation, context tracing, or profiling.…”

ALEA: Fine-Grain Energy Profiling with Basic Block Sampling

“…Other energy profiling tools build instruction-level power models bottom-up from gate-level models, or other hardware models extracted at design time to provide power profiles to simulators and prototyping environments [15], [16]. These inherently static models fail to capture the variability in instruction-level power consumption due to the context in which instructions execute in real programs.…”

“…Prior energy accounting tools can be broadly classified into two categories: Tools that measure energy by directly measuring power using on-board sensors or external instruments [3], [4], [5], [6], [7], [8]; and tools that model energy based on activity vectors of hardware performance counters, kernel event counters, finite state machines, or instruction counters in microbenchmarks [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19]. All of these tools can associate energy measurements with software contexts via manual instrumentation, context tracing, or profiling.…”

Data-Driven Optimization of Order Admission Policies in a Digital Print Factory

“…For system-wide design and optimization, time consumption prediction is crucial in almost any operational environment. For example, software optimization requires time-consumption estimation for each software stage [Tu et al 2014], yield optimization for embedded systems-on-chip requires time consumption information for each resource [Meyer et al 2014], and communication system design and optimization requires message response time [Schneider et al 2014]. Moreover, for real-time embedded systems, accurate estimates of completion times are preferred compared to simple worst-case execution times [Ding et al 2012;Ma et al 2013].…”