• 한국산업융합학회 논문집
• /
• 제24권1호
• /
• pp.69-77
• /
• 2021

The widely used low-cost design methodology for IoT devices is very popular. In such a networked device, memory is composed of flash memory, SRAM, DRAM, etc., and because it processes a large amount of data, memory design is an important factor for system performance. Therefore, each device selects optimized design factors such as function, performance and cost according to market demand. The design of a memory architecture available for low-cost IoT devices is very limited with the configuration of SRAM, flash memory, and DRAM. In order to process as much data as possible in the same space, an architecture that supports parallel processing units is usually provided. Such parallel architecture is a design method that provides high performance at low cost. However, it needs precise software techniques for instruction and data mapping on the parallel architecture. This paper proposes an instruction/data mapping method to support optimized parallel processing performance. The proposed method optimizes system performance by actively using hardware and software parallelism.

• JSTS:Journal of Semiconductor Technology and Science
• /
• 제15권4호
• /
• pp.477-483
• /
• 2015

The k-Winners-Take-All (kWTA) is an operation to find the largest k (>1) inputs among N inputs. Parallel search algorithm of kWTA for digital inputs is not invented yet, so most of digital kWTA architectures have O(N) time complexity. A parallel search algorithm for digital kWTA operation and the circuits for its VLSI implementation are presented in this paper. The proposed kWTA architecture can compare all inputs simultaneously in parallel. The time complexity of the new architecture is O(logN), so that it is scalable to a large number of digital data. The high-speed kWTA operation and its O(logN) dependency of the new architecture are verified by simulations. It takes 290 ns in searching for 5 winners among 1024 of 32 bit data, which is more than thousands of times faster than existing digital kWTA circuits, as well as existing analog kWTA circuits.

• 한국통신학회논문지
• /
• 제18권12호
• /
• pp.1902-1918
• /
• 1993

노이만 계산 모델의 병렬처리 구조는 구조 속성상의 취약성으로 인해 대량 병렬처리 구조로서는 한계가 있다. 데이터 플로우 계상 모델은 소프트웨어적 고 프로그램성과 하드웨어적 높은 개발 가능성을 갖고 있다. 그러나 실제 데이터 플로우 구조에서는 프로그래밍과 실험을 행하고자 할때, 노이만 방식의 기계는 많지만 실제 데이터 플로우 컴퓨터가 없으므로 대단히 어렵다. 본 논문에서는 일반적 재래 병렬처리기계중 하나인 이완결합 다중프로세서 시스템위에서 데이터 플로우 방식의 계산을 수행시킬 수 있는 프로그래밍 환경을 제시하였다. 에뮬레이터는 iPSC/2 하이퍼 큐프를 이용하여 Tagged Token 데이터 플로우 구조를 구축하였다. 본 에뮬레이터는 iPSC/2시스템에서 소프트웨어적 박층 실험으로 프로그래머의 입장에서는 iPSC/2 시스템이 데이터 플로우 주고로서 농작하는 것으로 간주한다. 여러 가지 수치 혹은 비수치 알고리즘을 데이터 플로우 어셈블리어로 구현하여 재래식 C 언어에 의한 것과 프로그램의 성능을 비교하였다. 이로써, 재래식 병렬처리 기계상의 에뮬레이터를 통한 실험적 데이터 플로우 계산을 행할 때 이 프로그래밍 환경의 효율성에 대하여도 검정하였다.

• 전기학회논문지
• /
• 제59권8호
• /
• pp.1489-1496
• /
• 2010

A feed forward differential architecture of analog PRML decoder is investigated to implement on analog parallel processing circuits. The conventional PRML decoder performs the trellis processing with the implementation of single stage in digital and its repeated use. The analog parallel processing-based PRML comes from the idea that the decoding of PRML is done mainly with the information of the first several number of stages. Shortening the trellis processing stages but implementing it with analog parallel circuits, several benefits including higher speed, no memory requirement and no A/D converter requirement are obtained. Most of the conventional analog parallel processing-based PRML decoders are differential architecture with the feedback of the previous decoded data. The architecture used in this paper is without feedback, where error metric accumulation is allowed to start from all the states of the decoding stage, which enables to be decoded without feedback. The circuit of the proposed architecture is simpler than that of the conventional analog parallel processing structure with the similar decoding performance. Characteristics of the feed forward differential architecture are investigated through various simulation studies.

• 전자공학회논문지B
• /
• 제33B권5호
• /
• pp.41-51
• /
• 1996

In this paper, anew parallel processing system based on a cluster architecture which provides scalability of a parallel processing system while maintains shared memory multiprocessor characteristics is proposed. In recent days low cost, high performnce microprocessors have led to construction of large scale parallel processing systems. Such parallel processing systems provides large scalability but are mainly used for scientific applications which have large data parallelism. A shared memory multiprocessor system like TICOM is currently used as aserver for the commercial application, however, the shared memory multiprocessor system is known to have very limited scalability. The proposed architecture can support scalability and performance of the parallel processing system while it provides adaptability for the commerical application, hence it can overcome the limitation of the shared memory multiprocessor. The architecture and characteristics of the proposed system shall be described. A proprietary hierarchical crsossbar network is designed for this system, of which the protocol, routing and switching technique and the signal transfer technique are optimized for the proposed architecture. The design trade-offs for the network are described in this paper and with simulation usihng the SES/workbench, it is explored that the network fits to the proposed architecture.

• 대한전자공학회논문지SD
• /
• 제40권7호
• /
• pp.518-525
• /
• 2003

본 논문은 리프팅 스킴(lifting scheme)의 분할 방법을 개선하여 고속 병렬 처리가 가능한 2차원 DWT(Discrete Wavelet Transform) 하드웨어 구조를 제안한다. 2차원 DWT 변환은 2차원 입력 데이터 전체에 대하여 연산이 수행되고 순차적으로 2차원 처리가 됨에 따라서 초기 및 전체 지연시간(latency)이 많이 걸린다. 본 논문에서는 처리속도와 지연 시간을 향상시키기 위해 개선된 분할 방법과 새로운 자원 공유 하드웨어 구조를 제안한다. 상호 연관성이 없는 데이터들을 4 개의 데이터 집합으로 분할하여 병렬 처리에 적합하도록 새로운 분할 방법을 제안하였다. 병렬처리 하드웨어 구조는 하드웨어의 자원 공유가 가능하도록 하기 위해 필터연산의 중간 값을 메모리에 저장할 수 있는 파이프라인 구조를 갖도록 설계하였다. 제안된 구조를 효율적으로 동작시킬 수 있도록 하드웨어 자원의 공유를 스케쥴링하여 초기지연과 전체지연 시간을 줄였다. 제안하는 구조는 기존의 병렬 처리 구조에 비해 초기 지연 및 전체 지연 시간을 각각 50%와 66%감소시키는 결과를 얻을 수 있었다.

• 전자공학회논문지B
• /
• 제32B권3호
• /
• pp.11-23
• /
• 1995

In this paper, a parallel pipelined processor model which acts as a small VLIW processor architecture and a scheduling algorithm for extracting instruction-level parallelism on this architecture are proposed. The proposed model has a dual-instruction mode which has maximum 4 basic operations being executed in parallel. By combining these basic operations, variable instruction set can be designed for various applications. The scheduling algorithm schedules basic operations for parallel execution and removes pipeline hazards by examining data dependency and resource conflict relations. In order to examine operation and evaluate the performance,a C compiler and a simulator are developed. By simulating various test programs with the compiler and the simulator, the characteristics and the performance result of the proposed architecture are measured.

• 한국시뮬레이션학회논문지
• /
• 제3권1호
• /
• pp.29-42
• /
• 1994

The architecture of a MIMD-type parallel computer system is specified: a simulator is developed to support design and evaluation of systems based on the architecture: and conducted with the simulator to evaluate system performance. The horizontal/vertical-bus(H/V-bus) system architecture provides an NxN array of processing elements which communicate with each other through a network of N horizontal buses and N vertical buses. The simulator, written in SLAM II and FORTRAN, is designed to provide high-resolution in simulating the IPC mechanism. Parameters provide the user with independent control of system size, PE speed and IPC mechanism speed. Results generated by the simulator include execution times, PE utilizations, queue lengths, and other data. The simulator is used to study system performance when a partial differential equation is solved by parallel Gauss-Seidel method. For comparisons, the benchmark is also executed on a single-bus system simulator that is derived from the H/V-bus system simulator. The benchmark is also solved on a single PE to obtain data for computing speedups. An extensive analysis of results is presented.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 2000년도 가을 학술발표논문집 Vol.27 No.2 (3)
• /
• pp.576-578
• /
• 2000

현재의 대부분의 VOD System에서는 Single Server System의 제약인 확장성과 안정적 서비스를 제공하기 위해 Multi-Server System을 사용하고 있다. Multiple Server에 Video Data를 Striping한 구조를 Parallel Video Server Architecture라 한다. 본 연구에서는 Parallel Video Server System 상에서 Data의 Striping Policy에 대해 고찰해보고 이때 발생하는 load balancing과 redundancy 문제의 해결책을 제시하였다. 또한 이를 실제 local Network 시스템에 적용하여 구현하였다.

• International Journal of Contents
• /
• 제1권1호
• /
• pp.50-58
• /
• 2005

Generally, multidimensional data such as image and spatial data require large amount of storage space. There is a limit to store and manage those large amounts of data in single workstation. If we manage the data on parallel computing environment which is being actively researched these days, we can get highly improved performance. In this paper, we propose a parallel multidimensional index structure that exploits the parallelism of the parallel computing environment. The proposed index structure is nP(processor)-nxmD(disk) architecture which is the hybrid type of nP-nD and 1P-nD. Its node structure in-creases fan-out and reduces the height of an index. Also, a range search algorithm that maximizes I/O parallelism is devised, and it is applied to k-nearest neighbor queries. Through various experiments, it is shown that the proposed method outperforms other parallel index structures.