A Performance Study of Zero Crossing Rate (ZCR) on Graphics Processors (GPUs) Using CUDA
Dept. of Electrical & Computing Engineering, Georgia Institute of Technology, Atlanta, GA, USA
Georgia Institute of Technology, 2012
@article{priyadarshi2012performance,
title={A Performance Study of Zero Crossing Rate (ZCR) on Graphics Processors (GPUs) Using CUDA},
author={Priyadarshi, Nydruva Biligere and Kim, Jongman},
year={2012}
}
The Ability to harness the power of the Graphics Processor Unit (GPU) enables us to show dramatic increases in computing performance using a parallel computing platform and programming model such as Nvidia CUDA. Compute Unified Device Architecture (CUDA) is NVIDIAs graphics programming API to perform General Purpose Graphics Processing Unit Programming (GPGPU). The General Purpose GPU Programming (GPGPU) environments have garnered universal access to computing performance that was only available to super-computers. This paper uses Nvidia CUDA v4.2 to explore the effectiveness of a GTX 480 GPU for a performance study of timbre (short-term feature) audio feature – Zero Crossing Rate (ZCR). The GPU Performance is compared to the implementations written using Vamp Plug-in and Sonic Annotator on an Intel Core i7 CPU 930@2.80GHz. While the CPU Implementations requires several seconds to perform an extraction, The GPU based approach has the capability to better this by 1.5X – 2.0X. The paper also discusses the feature extraction of other audio features in Future Work section & the advantages and inefficiencies of the CUDA Programming model.
April 6, 2013 by hgpu
