• 대한전자공학회논문지SD
• /
• 제44권10호
• /
• pp.23-32
• /
• 2007

본 논문에서는 3D 그래픽스(graphics), H.264/H.263/MPEG-4 같은 동영상 코덱, JPEG 혹은 JPEG2000 같은 정지영상 코덱, MP3 같은 오디오 코덱 등 다양한 멀티미디어 관련 기술을 효율적으로 구현하기 위한 재구성형 병렬 프로세서 구조가 제안된다. 제안된 구조는 메모리와 프로세서를 직접 연결하여 메모리 접근 시간과 소비전력를 감소시키고, 3D 그래픽스 처리 과정중 기하 단계의 부동소수점 연산을 지원한다. 또한 분할 SIMD(partitioned SIMD) 방식을 사용하여 하드웨어 비용을 줄이고, 명령어(instruction)의 조건부 실행(conditional execution)을 지원하여 알고리듬 개발이 용이하다.

• 대한전기학회논문지:전력기술부문A
• /
• 제48권5호
• /
• pp.636-642
• /
• 1999

The synthesis of the 3D images is the most important part of the virtual reality. The ray tracing is the best method for reality in the 3D graphics. But the ray tracing requires long computation time for the synthesis of the 3D images. So, we implement the ray tracing with software and hardware. Specially we design the hit-test unit with FPGA tool for the ray tracing. Hit-test unit is a very important part of ray tracing to improve the speed. In this paper, we proposed a new hit-test algorithm and apply the parallel architecture for hit-test unit to improve the speed. We optimized the arithmetic unit because the critical path of hit-test unit is in the multiplication part. We used the booth algorithm and the baugh-wooley algorithm to reduce the partial product and adapted the CSA and CLA to improve the efficiency of the partial product addition. Our new Ray tracing processor can produce the image about 512ms/F and can be adapted to real-time application with only 10 parallel processors.

• 조명전기설비학회논문지
• /
• 제20권6호
• /
• pp.104-110
• /
• 2006

본 논문에서는 인텔사의 StrongARM 프로세서에 내장형 리눅스 운영 시스템을 실장하여 내장형 웹 서버를 구현하고 ARM프로세서에 연결된 병렬포트의 입 출력을 HTTP 프로토콜을 이용하여 범용 웹 브라우저에 의하여 제어하는 초소형 웹 서버 시스템을 구현함을 다루었다. 이를 위하여 리눅스 운영 시스템의 HTTP를 실장하고 CGI에 의한 병렬포트 제어 프로그램을 구현하여 프로세서 보드의 메모리에 실장한다. 프로세서의 병렬포트에 입 출력을 제어하는 하드웨어 기능을 웹 서버와 브라우저를 이용하여 원격에서 제어할 수 있도록 구현하고 실험을 통하여 내장형 웹 서버의 구현을 보였다.

• 한국컴퓨터정보학회논문지
• /
• 제14권11호
• /
• pp.97-103
• /
• 2009

본 논문에서는 시스템의 변경이 많고 적은 비용으로 고성능 데이터 처리가 요구되는 응용분야에서 시스템의 유연성, 가격, 크기 및 성능을 개선하기 위한 목적으로 알테라(Altera)의 $Nios^{(R)}$ II 임베디드 프로세서(embedded processor) 4개를 사용하여 주종(master-slave)과 공유메모리(shared memory) 구조를 가지는 병렬처리 시스템을 설계하고 구현하였다. 설계한 병렬처리 시스템은 $Nios^{(R)}$ II 32bit RISC 프로세서. $SOPC^{(R)}$ Builder, $Quartus^{(R)}$ II, $ModelSim^{(R)}$으로 개발되었으며 설계한 병렬처리 시스템의 성능 평가는 $Terasic^{(R)}$사의 $DE2-70^{(R)}$ 레퍼런스 보드($Cyclone^{(R)}$ II(EP2C70F896C6N) FPGA)에서 검증하고 구현하였다. 설계한 병렬처리 시스템의 성능을 평가하기 위해서 1개, 2개, 4개의 프로세서로 512, 1,024, 2,048, 4,096, 8,192 N-point FFT(fast fourier transform) 연산을 수행하여 속도향상(Sp)과 시스템의 효율(Ep)을 평가하였다. 성능평가 결과 Sp는 1개의 프로세서를 사용한 경우에 비해서 2개의 프로세서를 사용한 경우 평균 1,8배, 4개의 프로세서를 사용한 경우에는 평균 2.4배의 속도향상을 보였다. 또한 Ep는 1개의 프로세서를 사용한 경우에는 1, 2개의 프로세서를 사용한 경우에는 평균 0.90, 4개의 프로세서를 사용한 경우에 평균 0.59를 보였다. 결과적으로 논문에서 구현된 병렬처리 시스템은 단일 프로세서를 사용하는 경우에 비해서 고성능 데이터 처리가 요구되는 분야에서 경제적인 시스템으로 구현할 수 있음을 보였다.

• 산업기술연구
• /
• 제21권A호
• /
• pp.73-80
• /
• 2001

Barrier is widely used for synchronization in parallel programs. Since the process arrived earlier than others should wait at the barrier, the total processor utilization decreases. In this paper, to find the sources of the barrier waiting time, parallel programs are executed on the various grain sizes through execution-driven simulations. In simulation studies, we found that even if approximately equal amounts of work are distributed to each processor, all processes may not arrive at a barrier at the same time. The reasons are that the different numbers of cache misses and instructions within partitioned grains result in the difference in arrival time of processors at the barrier.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2001년도 하계종합학술대회 논문집(2)
• /
• pp.269-272
• /
• 2001

In this paper we proposed the format converter design and implementation for real time image processing. This design method is based on realized the large processor-per-pixel array by integrated circuit technology in which this two types of integrated structure is can be classify associative parallel processor and parallel process with DRAM cell. Layout pitch of one-bit-wide logic is identical memory cell pitch to array high density PEs in integrate structure. This format converter design has control path implementation efficiently, and can be utilized the high technology without complicated controller hardware. Sequence of array instruction are generated by host computer before process start, and instructions are saved on unit controller. Host computer is executed the pixel-parallel operation starting at saved instructions after processing start

• International Journal of Computer Science & Network Security
• /
• 제21권12spc호
• /
• pp.570-578
• /
• 2021

Recently, in the field of engineering and scientific and technical calculations, problems of mathematical modeling, real-time problems, there has been a tendency towards rejection of sequential solutions for single-processor computers. Almost all modern application packages created in the above areas are focused on a parallel or distributed computing environment. This is primarily due to the ever-increasing requirements for the reliability of the results obtained and the accuracy of calculations, and hence the multiply increasing volumes of processed data [2,17,41]. In addition, new methods and algorithms for solving problems appear, the implementation of which on single-processor systems would be simply impossible due to increased requirements for the performance of the computing system. The ubiquity of various types of parallel systems also plays a positive role in this process. Simultaneously with the growing demand for parallel programs and the proliferation of multiprocessor, multicore and cluster technologies, the development of parallel programs is becoming more and more urgent, since program users want to make the most of the capabilities of their modern computing equipment[14,39]. The high complexity of the development of parallel programs, which often does not allow the efficient use of the capabilities of high-performance computers, is a generally accepted fact[23,31].

• 한국항해학회지
• /
• 제14권4호
• /
• pp.31-39
• /
• 1990

A systematic methodology for efficient implementation of processor arrays from regular iterative algorithms is proposed. One of the modern parallel processing array architectures is the Systolic arrays and we use it for processor arrays on this paper. On designing the systolic arrays, there are plenty of mapping functions which satisfy necessary conditions for its implementation to the time-space domain. In this paper, we sue a few conditions to reduce the total number of computable mapping functions efficiently. As a results of applying this methodology, efficient designs of systolic arrays could be done with considerable saving on design time and efforts.

• 한국컴퓨터정보학회논문지
• /
• 제18권9호
• /
• pp.1-10
• /
• 2013

본 논문에서는 대표적인 특징점 추출 알고리즘인 SIFT(Scale-Invariant Feature Transform)를 매니코어 프로세서를 이용하여 병렬 구현하고, 이를 실행 시간, 시스템 이용률, 에너지 효율 및 시스템 면적 효율 측면에서 분석하였다. 또한 기존의 고성능 CPU와 GPU(Graphics Processing Unit)와의 성능 비교를 통해 제안하는 매니코어의 잠재가능성을 입증하였다. 모의실험 결과, 매니코어를 이용한 SIFT 알고리즘 구현 결과는 기존의 OpenCV 구현 결과와 정확도면에서 동일하였고, 매니코어 구현은 고성능 CPU 및 GPU 구현보다 실행시간 측면에서 우수하였다. 또한 본 논문에서는 SIFT알고리즘의 옥타브 크기에 따른 에너지 효율 및 시스템 면적 효율을 분석하여 최적의 모델을 제시하였다.

• 전자통신동향분석
• /
• 제33권5호
• /
• pp.121-134
• /
• 2018

The Von Neumann based architecture of the modern computer has dominated the computing industry for the past 50 years, sparking the digital revolution and propelling us into today's information age. Recent research focus and market trends have shown significant effort toward the advancement and application of artificial intelligence technologies. Although artificial intelligence has been studied for decades since the Turing machine was first introduced, the field has recently emerged into the spotlight thanks to remarkable milestones such as AlexNet-CNN and Alpha-Go, whose neural-network based deep learning methods have achieved a ground-breaking performance superior to existing recognition, classification, and decision algorithms. Unprecedented results in a wide variety of applications (drones, autonomous driving, robots, stock markets, computer vision, voice, and so on) have signaled the beginning of a golden age for artificial intelligence after 40 years of relative dormancy. Algorithmic research continues to progress at a breath-taking pace as evidenced by the rate of new neural networks being announced. However, traditional Von Neumann based architectures have proven to be inadequate in terms of computation power, and inherently inefficient in their processing of vastly parallel computations, which is a characteristic of deep neural networks. Consequently, global conglomerates such as Intel, Huawei, and Google, as well as large domestic corporations and fabless companies are developing dedicated semiconductor chips customized for artificial intelligence computations. The AI Processor Research Laboratory at ETRI is focusing on the research and development of super low-power AI processor chips. In this article, we present the current trends in computation platform, parallel processing, AI processor, and super-threaded AI processor research being conducted at ETRI.