Name
Abstract | ||
|---|---|---|
Heterogeneous computing using Graphic Processing Units (GPUs) has become an attractive computing model given the available scale of data-parallel performance and programming standards such as OpenCL. However, given the energy issues present with GPUs, some devices can exhaust power budgets quickly. Better solutions are needed to effectively exploit the power efficiency available on heterogeneous systems. In this paper we evaluate the power-performance trade-offs of different heterogeneous signal processing applications. More specifically, we compare the performance of 7 different implementations of the Fast Fourier Transform algorithms. Our study covers discrete GPUs and shared memory GPUs (APUs) from AMD (Llano APUs and the Southern Islands GPU), Nvidia (Fermi) and Intel (Ivy Bridge). For this range of platforms, we characterize the different FFTs and identify the specific architectural features that most impact power consumption. Using the 7 FFT kernels, we obtain a 48% reduction in power consumption and up to a 58% improvement in performance across these different FFT implementations. These differences are also found to be target architecture dependent. The results of this study will help the signal processing community identify which class of FFTs are most appropriate for a given platform. More important, we have demonstrated that different algorithms implementing the same fundamental function (FFT) can perform vastly different based on the target hardware and associated programming optimizations. |
Year | DOI | Venue |
|---|---|---|
2013 | 10.1109/ISPASS.2013.6557174 | Performance Analysis of Systems and Software |
Keywords | Field | DocType |
fast Fourier transforms,graphics processing units,parallel architectures,power consumption,shared memory systems,AMD,FFT,Fermi,Intel,Ivy Bridge,Llano APU,Nvidia,OpenCL,Southern Islands GPU,computing model,data-parallel performance,discrete GPU,energy efficiency,energy issues,fast Fourier transform algorithm,graphic processing unit,heterogeneous computing,heterogeneous platform,heterogeneous signal processing application,heterogeneous system,power budget,power consumption,power efficiency,power-performance trade-off,programming optimization,programming standards,shared memory GPU,FFT,GPUs,OpenCL,Power | Electrical efficiency,Kernel (linear algebra),Signal processing,Shared memory,Ivy Bridge,Computer science,Efficient energy use,Parallel computing,Symmetric multiprocessor system,Fast Fourier transform | Conference |
ISBN | Citations | PageRank |
978-1-4673-5778-4 | 5 | 0.49 |
References | Authors | |
13 | 4 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Ukidave, Y. | 1 | 5 | 0.49 |
| Ziabari, A.K. | 2 | 5 | 0.49 |
| Mistry, P. | 3 | 20 | 1.94 |
| Schirner, G. | 4 | 5 | 0.49 |