DFT Performance Prediction in FFTW

Gu, Liang; Li, Xiaoming

doi:10.1007/978-3-642-13374-9_10

Cited by 1 publication

(4 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To improve library efficiency, several FFT performance models [3][4]7,[9][10] have been proposed to reduce the search time for an efficient FFT algorithm on a GPP.…”

Section: Fft Performance Modelmentioning

confidence: 99%

“…Gu et al [7] developed a performance model as a weighted linear combination of operations to reduce the search time in FFTW. Fraguela et al [4] modeled the memory stall time based on misses at different memory hierarchies, and could select the best performing DFT algorithms in SPIRAL.…”

Section: Fft Performance Modelmentioning

confidence: 99%

“…Fraguela et al [4] modeled the memory stall time based on misses at different memory hierarchies, and could select the best performing DFT algorithms in SPIRAL. The above models could efficiently estimate the actual runtime of N -point FFTs, and reduce the search time for optimal FFT implementations [3][4]7,[9][10] at post-silicon phase. However, they need data training to establish models, but do not establish obvious correlations between FFT performance and architecture parameters, such as register file size, memory bandwidth, and functional units.…”

Section: Fft Performance Modelmentioning

confidence: 99%

“…Due to the importance of FFT, several FFT performance models have been proposed to estimate the FFT runtime [5][6][7][8] and search for efficient FFT implementation [3][4]7,[9][10] . However, they do not comprehensively characterize the FFT performance impacts of various architecture parameters, thus are hard to be directly employed in GPP architecture optimization.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

An FFT Performance Model for Optimizing General-Purpose Processor Architecture

Chen

Liu

et al. 2011

J. Comput. Sci. Technol.

View full text Add to dashboard Cite

General-purpose processor (GPP) is an important platform for fast Fourier transform (FFT), due to its flexibility, reliability and practicality. FFT is a representative application intensive in both computation and memory access, optimizing the FFT performance of a GPP also benefits the performances of many other applications. To facilitate the analysis of FFT, this paper proposes a theoretical model of the FFT processing. The model gives out a tight lower bound of the runtime of FFT on a GPP, and guides the architecture optimization for GPP as well. Based on the model, two theorems on optimization of architecture parameters are deduced, which refer to the lower bounds of register number and memory bandwidth. Experimental results on different processor architectures (including Intel Core i7 and Godson-3B) validate the performance model.The above investigations were adopted in the development of Godson-3B, which is an industrial GPP. The optimization techniques deduced from our performance model improve the FFT performance by about 40%, while incurring only 0.8% additional area cost. Consequently, Godson-3B solves the 1024-point single-precision complex FFT in 0.368 µs with about 40 Watt power consumption, and has the highest performance-per-watt in complex FFT among processors as far as we know. This work could benefit optimization of other GPPs as well.

show abstract

“…To improve library efficiency, several FFT performance models [3][4]7,[9][10] have been proposed to reduce the search time for an efficient FFT algorithm on a GPP.…”

Section: Fft Performance Modelmentioning

confidence: 99%

Section: Fft Performance Modelmentioning

confidence: 99%