• Proceedings of the Korean Information Science Society Conference
• /
• 2005.11a
• /
• pp.985-987
• /
• 2005

자료의 크기가 방대한 응용 프로그램에서는 메인 메모리와 저장 장치간의 자료 입출력(I/O)이 전체 계산의 주요 병목 요인으로 작용한다. 본 논문은 [2004 한국정보처리학회 추계논문집 제11권 제2호 1139-1142]에서 제안된 그리드(grid) 그래프를 위한 입출력 효율적인 depth-first search (DFS) 알고리즘을 지원하기 위한 입출력 효율적인 영역 구분자를 구하는 알고리즘을 제안 한다. 그 입출력 복잡도 (I/O-Complexity)는 O(Soft(N))이다. 여기서 N=|V|+|E|이고 $sort(N)=\Theta((N/B)log_{M/B}(N/B))$이다.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2003.06a
• /
• pp.118-121
• /
• 2003

본 논문은 동영상 압축 부호화에 대한 표준안인 MPEG기반의 압축 비디오 stream에서［1, 2］, 질의 영상에 대한 효율적인 검색 기법을 제안한다. 비디오 검색은 높은 차원의 색인 정보를 이용하는데, 높은 차원의 data set을 색인 정보로 하여 효율적인 검색 능력을 보여주는 KD-Trees(K Dimensional-Trees)알고리즘［3］을 비디오 검색기법에 적용하고자 한다. 먼저, key frame에 PCA (Principal Component Analysis) 알고리즘［4］을 이용하여 색인 정보를 추출한 다음, 추출된 색인 정보를 KD-Trees에 적용하여 효율적인 검색을 가능하게 한다. 실험 결과, 기존의 검색 기법보다 상당한 양의 처리 시간과 메모리 공간을 줄일 수 있음을 보였다.

• Journal of Broadcast Engineering
• /
• v.27 no.4
• /
• pp.561-568
• /
• 2022

Most deep learning-based video object segmentation methods perform the segmentation with past prediction information stored in external memory. In general, the more past information is stored in the memory, the better results can be obtained by accumulating evidence for various changes in the objects of interest. However, all information cannot be stored in the memory due to hardware limitations, resulting in performance degradation. In this paper, we propose a method of storing new information in the external memory without additional memory allocation. Specifically, after calculating the attention score between the existing memory and the information to be newly stored, new information is added to the corresponding memory according to each score. In this way, the method works robustly because the attention mechanism reflects the object changes well without using additional memory. In addition, the update rate is adaptively determined according to the accumulated number of matches in the memory so that the frequently updated samples store more information to maintain reliable information.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.43 no.5 s.311
• /
• pp.8-16
• /
• 2006

This paper proposes and implements consistency algorithm that is serviced streaming data efficiently in VOD system. The media data is loaded into each node by Round Robin method. The barrier mechanism is changed into the minimum data factor(SH, GOP) of media data in this paper. In addition in order to fast media data service, the additional features are implemented in the consistency algorithm. Additional feature of the consistency algorithm is added. First, time synchronization algorithm is added the suggested consistency algorithm. Second, the prepaging mechanism supports efficient buffering service. I experimented the suggested consistency algorithm two DSM systems. The performance is good due to the last barrier algorithm.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.45 no.9
• /
• pp.33-40
• /
• 2008

As the speed of memory is improved vey fast the advanced test equipments are needed to test the ultra-high speed memory devices efficiently. It is necessary to develop the Algorithmic Pattern Generator (ALPG) that tests fast memory devices effectively using the instructions that testers want to use. In this paper, we propose a new parallel ALPG for the ultra-high speed memory testing. The proposed ALPG can generate patterns for fast memory devices at high speed using manual instructions by the Instruction Analyzer.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.5
• /
• pp.83-93
• /
• 2013

Recently, the SURF algorithm, which is conjugated for object tracking system as part of many computer vision applications, is a well-known scale- and rotation-invariant feature detection algorithm. The SURF, due to its high computational complexity, there is essential to develop a hardware accelerator in order to be used on an IP in embedded environment. However, the SURF requires a huge local memory, causing many problems that increase the chip size and decrease the value of IP in ASIC and SoC system design. In this paper, we proposed a way to design a SURF algorithm in hardware with greatly reduced local memory by partitioning the algorithms into several Sub-IPs using external memory and a DMA. To justify validity of the proposed method, we developed an example of simplified object tracking algorithm. The execution speed of the hardware IP was about 31 frame/sec, the logic size was about 74Kgate in the 30nm technology with 81Kbytes local memory in the embedded system platform consisting of ARM Cortex-M0 processor, AMBA bus(AHB-lite and APB), DMA and a SDRAM controller. Hence, it can be used to the hardware IP of SoC Chip. If the image processing algorithm akin to SURF is applied to the method proposed in this paper, it is expected that it can implement an efficient hardware design for target application.

• Journal of KIISE:Information Networking
• /
• v.36 no.1
• /
• pp.57-67
• /
• 2009

Packet forwarding in the Internet routers is to find out the longest prefix that matches the destination address of an input packet and to forward the input packet to the output port designated by the longest matched prefix. The IP address lookup is the key of the packet forwarding, and it is required to have efficient data structures and search algorithms to provide the high-speed lookup performance. In this paper, an efficient IP address lookup algorithm using binary search is investigated. Most of the existing binary search algorithms are not efficient in search performance since they do not provide a balanced search. The proposed binary search algorithm performs perfectly balanced binary search using switch pointers. The performance of the proposed algorithm is evaluated using actual backbone routing data and it is shown that the proposed algorithm provides very good search performance without increasing the memory amount storing the forwarding table. The proposed algorithm also provides very good scalability since it can be easily extended for multi-way search and for large forwarding tables

• Journal of IKEEE
• /
• v.24 no.2
• /
• pp.559-569
• /
• 2020

One of the deep-running algorithms, CNN's artificial intelligence application uses off-chip memory to store data on the Convolution Layer. DMA can reduce processor load at every data transfer. It can also reduce application performance degradation by varying the order in which data from the Convolution layer is transmitted to the global buffer of the accelerator. For basic layouts with continuous memory addresses, SG-DMA showed about 3.4 times performance improvement in pre-setting DMA compared to using ordinaly DMA, and for Ideal layouts with discontinuous memory addresses, the ordinal DMA was about 1396 cycles faster than SG-DMA. Experiments have shown that a combination of memory data layout and DMA can reduce the DMA preset load by about 86 percent.

• Proceedings of the Korean Information Science Society Conference
• /
• 2012.06a
• /
• pp.349-351
• /
• 2012

낸드 플래시 메모리에서 가비지 콜렉션은 블록의 유효한 데이터들을 새로운 블록으로 옮기고 오래된 블록을 지우는 과정이다. 가비지 콜렉션에 의해 옮겨지는 페이지들은 작업의 양과 형태에 따라 오랫동안 유효한 페이지로 존재하기도 하고 그렇지 않은 경우도 있다. 본 논문에서는 반도체 디스크(Solid State Drive, 이하 SSD)에서 가비지 콜렉션이 비효율적으로 일어나는 경우를 정의하고 비효율적 가비지 콜렉션 과정으로 인한 비용을 줄이는 캐시 방출 기법을 소개한다. 이 기법을 시뮬레이션 해본 결과 작업 형태가 순차적일 때 LRU 캐시 알고리즘과 같이 사용되면 가비지 콜렉션에 의해 옮겨지는 페이지를 12%, 전체 쓰기 연산 횟수를 9%까지 줄일 수 있었고 블록 단위 LRU 알고리즘과 사용했을 때도 보다 좋은 성능을 보였다.

• Journal of the Korea Society of Computer and Information
• /
• v.15 no.6
• /
• pp.57-64
• /
• 2010

Due to the limited storage space in video proxy server, it is often required to replace the old video data which is not serviced for long time with the newly requested video. This replacement causes the service delay and increase of network traffic. To circumvent this problem, we propose the an efficient replacement scheme in a video proxy server. In addition, we present a video data management technique for decreasing the number of replacement in video proxy server. For this purpose, we employ a memory in video proxy server. If the video segment which is loaded in memory is requested once again by a different user, this segment is resided in memory. The video in the memory is stored in the video proxy server depending on the consuming pattern by users. The simulation results show that the proposed algorithm performs better than other algorithms in terms of packet hit rate and number of packet replacement.