• Proceedings of the Korean Society of Computer Information Conference
• /
• 2019.07a
• /
• pp.97-98
• /
• 2019

본 논문에서는 대용량의 방송 콘텐츠를 전송하기 위해 다수개의 채널을 결합하여 안정적이고 고속으로 전송하기 위한 방송콘텐츠를 전송하기 위한 테스트베드를 제안한다. 제안하는 테스트베드의 첫 번째 목적은 하나의 방송채널 용량을 초과하는 대용량 방송 콘텐츠를 다수개의 채널을 결합하여 전송하기 위한 것이며, 두 번째 목적은 다채널로 입력된 데이터를 다양한 방법의 병렬 알고리즘을 적용하여 FPGA에 적용한 후 그 결과를 테스트하기 위한 것이다. 이를 위하여 제안하는 테스트베드는 다채널을 위한 입력 보드와 전반적인 제어를 위한 CPU 보드, 병렬 알고리즘 등을 테스트하기 위한 FPGA 보드, 그리고 3개의 보드들을 연결하기 위한 베이스 보드로 구성되었다. 제안하는 테스트베드 환경에서 다채널 대용량의 데이터를 병렬처리 할 수 있는 병렬 알고리즘들을 지속적으로 개발하고 테스트하여 다채널 대용량의 실시간 처리가 가능한 영상처리 시스템을 개발하는 것이다.

• International Journal of Computer Science & Network Security
• /
• v.22 no.2
• /
• pp.193-198
• /
• 2022

Face Detection is one of the most important aspects of image processing, it considers a time-consuming problem in real-time applications such as surveillance systems, face recognition systems, attendance system and many. At present, commodity hardware is getting more and more heterogeneity in terms of architectures such as GPU and MIC co-processors. Utilizing those co-processors along with the existing traditional CPUs gives the algorithm a better chance to make use of both architectures to achieve faster implementations. This paper presents a hybrid implementation of the face detection based on the local binary pattern (LBP) algorithm that is deployed on both traditional CPU and MIC co-processor to enhance the speed of the LBP algorithm. The experimental results show that the proposed implementation achieved improvement in speed by 3X when compared to a single architecture individually.

• 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.

• Journal of KIISE:Computer Systems and Theory
• /
• v.27 no.4
• /
• pp.401-407
• /
• 2000

Modern microprocessors exploit instruction-level parallel processing to increase the performance. Especially VLIW processors supported by the parallelizing compiler are used more and more in specific applications such as high-end DSP and graphic processing. Bus-based VLIW architecture was proposed for these specific applications and it was designed to reduce the overhead of forwarding unit and the instruction width. In this paper, a optimizing scheduling compiler developed for the proposed bus-based VLIW processor is introduced. First, the method to model interconnections between buses and resource usage patterns is described. Then, on the basis of the modeling, machine-dependent optimization techniques such as bus-to-register promotion, copy coalescing and operand substitution were implemented. Optimization techniques for general-purpose VLIW microprocessors such as selective scheduling and enhanced pipelining scheduling(EPS) were also implemented. The experiment result shows about 20% performance gain for multimedia application benchmarks.

• KIPS Transactions on Computer and Communication Systems
• /
• v.6 no.12
• /
• pp.469-478
• /
• 2017

The mutual exclusion is originally based on the theory of race condition prevention in symmetric multi-processor operating systems. But recently, due to the generalization of multi-core processors, its application range has been rapidly shifted to parallel processing application domain. POSIX thread, WIN32 thread, and Java thread, which are typical parallel processing application development environments, provide a unique mutual exclusion mechanism for each of them. Applications that are very sensitive to performance in these environments may want to reduce the burden of mutual exclusion, even at some cost, such as inconvenience of coding. In this study, we implement Dekker's and Peterson's algorithm in the form of busy-wait and processor-yield in various platforms, and compare the performance of them with the built-in mutual exclusion mechanisms to evaluate the usability of the classic algorithms. The analysis result shows that Dekker's algorithm of processor-yield type is superior to the built-in mechanisms in POSIX and WIN32 thread environments at least 2 times and up to 70 times, and confirms that the practicality of the algorithm is sufficient.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.49 no.3
• /
• pp.8-14
• /
• 2012

Improving the speed of image processing is in great demand according to spread of high quality visual media or massive image applications such as 3D TV or movies, AR(Augmented reality). SIMD computer attached to a host computer can accelerate various image processing and massive data operations. MAMS is a multi-access memory system which is, along with multiple processing elements(PEs), adequate for establishing a high performance pipelined SIMD machine. MAMS supports simultaneous access to pq data elements within a horizontal, a vertical, or a block subarray with a constant interval in an arbitrary position in an $M{\times}N$ array of data elements, where the number of memory modules(MMs), m, is a prime number greater than pq. MAMS-PP4 is the first realization of the MAMS architecture, which consists of four PEs in a single chip and five MMs. This paper presents implementation of image processing algorithms and performance analysis for MAMS-PP16 which consists of 16 PEs with 17 MMs in an extension or the prior work, MAMS-PP4. The newly designed MAMS-PP16 has a 64 bit instruction format and application specific instruction set. The author develops a simulator of the MAMS-PP16 system, which implemented algorithms can be executed on. Performance analysis has done with this simulator executing implemented algorithms of processing images. The result of performance analysis verifies consistent response of MAMS-PP16 through the pyramid operation in image processing algorithms comparing with a Pentium-based serial processor. Executing the pyramid operation in MAMS-PP16 results in consistent response of processing time while randomly response time in a serial processor.

• KIPS Transactions on Computer and Communication Systems
• /
• v.11 no.6
• /
• pp.167-174
• /
• 2022

PIPO lightweight block ciphers were announced in ICISC'20. In this paper, a single-block optimization implementation and parallel optimization implementation of PIPO lightweight block cipher ECB, CBC, and CTR operation modes are performed on a 32-bit RISC-V processor. A single block implementation proposes an efficient 8-bit unit of Rlayer function implementation on a 32-bit register. In a parallel implementation, internal alignment of registers for parallel implementation is performed, and a method for four different blocks to perform Rlayer function operations on one register is described. In addition, since it is difficult to apply the parallel implementation technique to the encryption process in the parallel implementation of the CBC operation mode, it is proposed to apply the parallel implementation technique in the decryption process. In parallel implementation of the CTR operation mode, an extended initialization vector is used to propose a register internal alignment omission technique. This paper shows that the parallel implementation technique is applicable to several block cipher operation modes. As a result, it is confirmed that the performance improvement is 1.7 times in a single-block implementation and 1.89 times in a parallel implementation compared to the performance of the existing research implementation that includes the key schedule process in the ECB operation mode.

• Journal of the Korea Convergence Society
• /
• v.12 no.5
• /
• pp.23-29
• /
• 2021

Recent deep learning-based natural language processing studies are conducting research to improve performance by training large amounts of data from various sources together. However, there is a possibility that the methodology of learning by combining data from various sources into one may prevent performance improvement. In the case of machine translation, data deviation occurs due to differences in translation(liberal, literal), style(colloquial, written, formal, etc.), domains, etc. Combining these corpora into one for learning can adversely affect performance. In this paper, we propose a new Corpus Weight Balance(CWB) method that considers the balance between parallel corpora in machine translation. As a result of the experiment, the model trained with balanced corpus showed better performance than the existing model. In addition, we propose an additional corpus construction process that enables coexistence with the human translation market, which can build high-quality parallel corpus even with a monolingual corpus.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.46 no.3
• /
• pp.96-106
• /
• 2009

Extending the scale of the distributed processing in a single Ethernet cluster is physically restricted by maximum ports per switch. This paper presents an implementation of MPI-based multicluster consisting of multiple Ethernet switches for extending the scale of distributed processing, and a asymptotical analysis for communication overhead through execution-time analysis model. To determine an optimum task partitioning, we analyzed the processing time for various partitioning schemes, and AAP(accumulator array partitioning) scheme was finally chosen to minimize the overall communication overhead. The scope of data partitioned in AAP was modified to fit for incremented nodes, and suitable load balancing algorithm was implemented. We tried to alleviate the communication overhead through exploiting the pipelined broadcast and flat-tree based result gathering, and overlapping of the communication and the computation time. We used the linear pipeline broadcast to reduce the communication overhead in intercluster which is interconnected by a single link. Experimental results shows nearly linear speedup by the proposed parallel distributed GHT implemented on MPI-based Ethernet multicluster with four 100Mbps Ethernet switches and up to 128 nodes of Pentium PC.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.16 no.4
• /
• pp.2749-2756
• /
• 2015

Recently, Depending on the quality of data increases, the problem of time-consuming to process the image is raised by being required to accelerate the image processing algorithms, in a traditional CPU and CUDA(Compute Unified Device Architecture) based recognition system for computing speed and performance gains compared to OpenMP When character recognition has been learned by the system to measure the input by the character data matching is implemented in an environment that recognizes the region of the well, so that the font of the characters image learning English alphabet are each constant and standardized in size and character an image matching method for calculating the matching has also been implemented. GPGPU (General Purpose GPU) programming platform technology when using the CUDA computing techniques to recognize and use the four cores of Intel i5 2500 with OpenMP to deal quickly and efficiently an algorithm, than the performance of existing CPU does not produce the rate of four times due to the delay of the data of the partition and merge operation proposed a method of improving the rate of speed of about 3.2 times, and the parallel processing of the video card that processes a result, the sequential operation of the process compared to CPU-based who performed the performance gain is about 21 tiems improvement in was confirmed.