• Journal of the Institute of Convergence Signal Processing
• /
• v.16 no.4
• /
• pp.139-148
• /
• 2015

In this paper, the optimum coding method is proposed by using open library in order to improve the performance of a software correlator developed for Korea-Japan Joint VLBI Correlator(KJJVC). The correlation system for VLBI observing system is generally implemented with hardware using ASIC or FPGA because the computational quantity is increased geometrically according to the participated observatory number. However, the software correlation system is recently constructed at a massive server such as a cluster using software according to the development of computing power. Since VLBI correlator implemented with hardware is able to conduct data processing with real-time or quasi real-time compared with mostly observational time, software correlation has to perform optimal data processing in coding work so as to have the same performance as that of the hardware. Therefore, in this paper, the experimental comparison was conducted by open-source based fftw library released in FFT processing stage, which is the most important part of the correlator system for performing optimum coding work in software development phase, such as general method using fftw library or methods using SSE(Streaming SIMD Extensions), shared memory, or OpenMP, and method using merged techniques listed above. Through the experimental results, the proposed optimum coding method for improving the performance of developed software correlator using fftw library, shared memory and OpenMP is effectively confirmed by reducing correlation time compared with conventional method.

• ETRI Journal
• /
• v.45 no.6
• /
• pp.1035-1045
• /
• 2023

We propose a novel graphics processing unit (GPU) algorithm that can handle a large-scale 3D fast Fourier transform (i.e., 3D-FFT) problem whose data size is larger than the GPU's memory. A 1D FFT-based 3D-FFT computational approach is used to solve the limited device memory issue. Moreover, to reduce the communication overhead between the CPU and GPU, we propose a 3D data-transposition method that converts the target 1D vector into a contiguous memory layout and improves data transfer efficiency. The transposed data are communicated between the host and device memories efficiently through the pinned buffer and multiple streams. We apply our method to various large-scale benchmarks and compare its performance with the state-of-the-art multicore CPU FFT library (i.e., fastest Fourier transform in the West [FFTW]) and a prior GPU-based 3D-FFT algorithm. Our method achieves a higher performance (up to 2.89 times) than FFTW; it yields more performance gaps as the data size increases. The performance of the prior GPU algorithm decreases considerably in massive-scale problems, whereas our method's performance is stable.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.07a
• /
• pp.563-566
• /
• 2020

본 논문은 초고해상도 컴퓨터 홀로그램 생성을 위한 GPU 기반 2D Shift-FFT 의 효율적인 구현 방법을 제안한다. 본 연구가 제안하는 알고리즘은 기존에 여섯 단계로 이루어진 처리과정을 다섯 단계로 줄임으로서, 병렬처리에서 비효율적인 메모리 접근 과정을 줄인다. 또한, 핀드(pinned) 메모리 기반의 CPU-GPU 데이터 통신 통로인 핀드 버퍼(pinned buffer)를 사용하고 다중 스트림을 채용함으로써, GPU 활용의 주요 병목원인이 되는 데이터 통신의 부하를 줄이고 GPU 활용 효율을 높인다. 본 연구는 제안하는 알고리즘의 효용성을 증명하기 위해 서로 다른 두 시스템에 알고리즘을 구현하고, 다양한 크기의 행렬에 대한 2D-FFT 처리에 대한 성능을 측정하였다. 그 결과, CPU 기반의 FFTW 라이브러리 대비 최대 3 배, 동일한 GPU 를 사용하는 cuFFT 라이브러리 대비 최대 1.5 배 높은 성능을 달성하였다. 이러한 결과는, 본 연구가 제안하는 알고리즘의 효용성을 보여주는 결과다.

• Journal of Korea Multimedia Society
• /
• v.13 no.7
• /
• pp.960-972
• /
• 2010

In this paper, we present a software for placing instruments on stage based on the acoustic properties of the concert hall. In order to simulate the changes in sound depending on the positions of the instruments, we incorporated the idea of location-based reverberation effect which can be realized through the convolution of instrument sounds with the impulse responses from the respective instrument positions. And we developed a software with a real-time convolution engine which enables the user to conveniently simulate the resulting sound of various instrument placements. The software was tested with the impulse response data measured at two concert halls of the National Center for Korean Traditional Performing Arts and Korean traditional instrument sounds. Results of these experiments show that simulated reverberation effects properly represent the spatial placement of instruments on stage.

• Journal of the Korea Convergence Society
• /
• v.7 no.5
• /
• pp.15-21
• /
• 2016

Recently the degree of integration of processor and developed rapidly. However, clock speed is not increased, a situation that increases the number of cores in the processor. In this paper, we analyze the performance of a typical Intel Xeon Phi of many core process used for the current operation accelerate. Utilizing the Quantum ESPRESSO, which was calculated using the FFTW library. By varying the number of ranks in MPI when running the benchmarks the performance Xeon Phi. The result shows a good performance in the handling of four job on one physical core. However, four or more to expand the number of MPI Rank is degraded. Through this convergence it was found to improve the performance of Quantum ESPRESSO. It is possible to check the hardware characteristics of the Xeon Phi.

• Journal of the Institute of Convergence Signal Processing
• /
• v.16 no.4
• /
• pp.175-181
• /
• 2015

In this paper, we propose a design method for a digital filter using software in order to analyze the radio astronomy observation data. Recently the analysis method for radio astronomy observing system is transferring from hardware to software by developing of state-of-the-art of computer system. The existing hardware system is not able to easily change the specification because it is implemented to meet special requirements and it takes a high cost and time. In case of software, however, it has an advantage to implement with small cost if open software is used, and flexibly changes to satisfy the desired specification. But, in order to analyze the massive data like radio astronomy with software, the good performance system is needed for computer. Therefore, this paper proposes a digital filter design method using software with the same performance as that of digital filter implemented with hardware in observation system which is operated by the KVN(Korean VLBI Network). To design a digital filter, the proposed method is performed with standard C language and the simulation is conducted with GNU(GNU's Not Unix) Octave and investigated to show its effectiveness. In addition, for the high speed operation of the designed digital filter, the SSE(Streaming SIMD Extensions) library is adopted for available parallel operation. By the proposed digital filter, the digital filtering is performed for the wide band observation data in the KVN observation mode, the filtering result of narrow band observation has no ripple inside of stop band, and confirmed the effectiveness of the proposed method.