Improving Mobile SOC's Performance as an Energy Efficient DSP Platform with Heterogeneous Computing

Mobile system-on-chip (SOC) technology is improving at a staggering rate spurred primarily by the adoption of smartphones and tablets. This rapid innovation has allowed the mobile SOC to be considered in everything from high performance computing to embedded applications. In this work, modern SOC's heterogeneous computing capabilities are evaluated with a focus toward digital signal processing (DSP). Evaluation is conducted on modern consumer devices running Android operating system and leveraging the relatively new RenderScript Compute to utilize CPU resources alongside other compute resources such as graphics processing units (GPUs) and digital signal processors. In order to benchmark these concepts, several implementations of both the discrete Fourier transform (DFT) and the fast Fourier transform (FFT) are tested across devices. The results show both improvement in performance and energy efficiency on many devices compared to traditional Java implementations and indicate that the mobile SOC is a relevant platform for DSP applications.

Name thesis:degree_name

Computer Engineering

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Electrical and Computer Engineering

Year

2015

Author dc:creator

• Briggs, Matthew

Contributors dc:contributor

• Zarkesh-Ha, Payman
• Plusquellic, James
• Pollard, L. Howard

• SOC
• Android
• Renderscript
• DFT
• FFT
• Power

Language dc:language

English

Repository record dc:identifier

https://digitalrepository.unm.edu/ece_etds/39

OAI identifier oai:identifier

oai:digitalrepository.unm.edu:ece_etds-1038