• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2020년도 추계학술발표대회
• /
• pp.12-15
• /
• 2020

Data access characteristics can directly affect the efficiency of the system execution. This research is to design an accurate predictor by using historical memory access information, where highly accessible data can be migrated from low-speed storage (SSD/HHD) to high-speed memory (Memory/CPU Cache) in advance, thereby reducing data access latency and further improving overall performance. For this goal, we design a locally weighted linear regression prefetch scheme to cope with irregular access patterns in large graph processing applications for a DARM-PCM hybrid memory structure. By analyzing the testing result, the appropriate structural parameters can be selected, which greatly improves the cache prefetching performance, resulting in overall performance improvement.

• 전자공학회논문지
• /
• 제52권5호
• /
• pp.66-73
• /
• 2015

OFDM 기반 통신시스템에서 IFFT는 중요한 핵심 컴포넌트 중의 하나이다. 본 논문에서는 OFDM 기반 통신시스템을 위한 메모리가 효율적인 새로운 IFFT 설계 방법을 제안한다. OFDM 기반 통신시스템에서 사용되는 IFFT의 입력신호는 데이터 변조신호, 파일럿과 널(null) 신호로 구성된다. 제안한 방법은 IFFT 입력신호의 매핑을 통해 IFFT에서 가장 큰 메모리를 차지하는 비트리버스의 메모리를 감소시키는 데 초점을 둔다. 비트리버스의 메모리 크기를 감소시키기 위해 DIT기반 구조에 적합한 선택 매핑기법을 제안한다. 시뮬레이션을 통해 제안한 방법이 기존 방법과 비교하여 약 50%의 메모리가 감소됨을 보인다.

• 한국인터넷방송통신학회논문지
• /
• 제20권1호
• /
• pp.149-153
• /
• 2020

본 논문은 스마트폰 시스템에서 DRAM과 NVRAM으로 구성된 이기종 메모리를 위한 페이지 배치 기법을 제안한다. 이기종 메모리에 관한 기존 연구와 달리 본 논문은 메모리 접근에 대한 오프라인 분석에 기반하여 메모리 페이지를 배치한다. 이는 스마트폰 메모리 접근이 애플리케이션의 종류와 무관하게 특정 주소 영역에 집중적으로 나타나며, 쓰기 연산에 있어 그 편향성이 일관되게 나타난다는 점을 반영한 것이다. 제안한 기법은 오프라인 분석 결과를 토대로 NVRAM에 쓰기 트래픽이 적게 발생하도록 페이지 배치를 수행하며, 실험 결과 NVRAM에 발생하는 쓰기량을 성능 저하 없이 평균 61% 줄이는 것을 확인할 수 있었다.

• JSTS:Journal of Semiconductor Technology and Science
• /
• 제5권4호
• /
• pp.262-269
• /
• 2005

Semiconductor scientists and engineers ideally desire the faster but the cheaper non-volatile memory devices. In practice, no single device satisfies this desire because a faster device is expensive and a cheaper is slow. Therefore, in this paper, we use heterogeneous non-volatile memories and construct an efficient hierarchy for them. First, a small RAM device (e.g., MRAM, FRAM, and PRAM) is used as a write buffer of flash memory devices. Since the buffer is faster and does not have an erase operation, write can be done quickly in the buffer, making the write latency short. Also, if a write is requested to a data stored in the buffer, the write is directly processed in the buffer, reducing one write operation to flash storages. Second, we use many types of flash memories (e.g., SLC and MLC flash memories) in order to reduce the overall storage cost. Specifically, write requests are classified into two types, hot and cold, where hot data is vulnerable to be modified in the near future. Only hot data is stored in the faster SLC flash, while the cold is kept in slower MLC flash or NOR flash. The evaluation results show that the proposed hierarchy is effective at improving the access time of flash memory storages in a cost-effective manner thanks to the locality in memory accesses.

• 반도체디스플레이기술학회지
• /
• 제22권2호
• /
• pp.6-10
• /
• 2023

High-potential data can be predicted and stored in the cache to prevent cache misses, thus reducing the processor's request and wait times. As a result, the processor can work non-stop, hiding memory latency. By utilizing the temporal/spatial locality of memory access, the prefetcher introduced to improve the performance of these computers predicts the following memory address will be accessed. We propose a prefetcher that applies the GRU model, which is advantageous for handling time series data. Display the currently accessed address in binary and use it as training data to train the Gated Recurrent Unit model based on the difference (delta) between consecutive memory accesses. Finally, using a GRU model with learned memory access patterns, the proposed data prefetcher predicts the memory address to be accessed next. We have compared the model with the multi-layer perceptron, but our prefetcher showed better results than the Multi-Layer Perceptron.

• 한국정보통신학회논문지
• /
• 제14권11호
• /
• pp.2541-2547
• /
• 2010

본 논문에서는 대기 상태에서 저전력 eFuse OTP 메모리 IP틀 구현하기 위해 속도가 문제가 되지 않는 반복되는 블록 회로에서 1.2V 로직 트랜지스터 대신 누설 (off-leakage) 전류가작은 3.3V의 MV (Medium Voltage) 트랜지스터로 대체하는 설계기술을 제안하였다. 그리고 읽기 모드에서 RWL (Read Word-Line)과 BL의 기생하는 커패시턴스를 줄여 동작전류 소모를 줄이는 듀얼 포트 (Dual-Port) eFuse 셀을 사용하였다. 프로그램 전압에 대한 eFuse에 인가되는 프로그램 파워를 모의실험하기 위한 등가회로를 제안하였다. 하이닉스 90나노 CMOS 이미지 센서 공정을 이용하여 설계된 512비트 eFuse OTP 메모리 IP의 레이아웃 크기는 $342{\mu}m{\times}236{\mu}m$이며, 5V의 프로그램 전압에서 42개의 샘플을 측정한 결과 프로그램 수율은 97.6%로 양호한 특성을 얻었다. 그리고 최소 동작 전원 전압은 0.9V로 양호하게 측정되었다.

• JSTS:Journal of Semiconductor Technology and Science
• /
• 제2권2호
• /
• pp.111-124
• /
• 2002

The ferroelectric RAM (FeRAM) has a great advantage for a system on a chip (SOC) and mobile product memory, since FeRAM not only supports non-volatility but also delivers a fast memory access similar to that of DRAM and SRAM. This work develops at three levels: 1) low voltage operation with boost voltage control of bitline and plateline, 2) reducing bitline capacitance with multiple divided sub cell array, and 3) increasing chip performance with write operation sharing both active and precharge time period. The key techniques are implemented on the proposed hierarchy bitline scheme with proposed hybrid-bitline and high voltage boost control. The test chip and simulation results show the performance of sub-1.5 voltage operation with single step pumping voltage and self-boost control in a cell array block of 1024 ($64{\;}{\times}{\;}16$) rows and 64 columns.

• 한국시뮬레이션학회논문지
• /
• 제11권3호
• /
• pp.49-58
• /
• 2002

It takes 2 or 3 months to manufacture memory device. Defect has to exist owing to hundreds of processes. If there are too many defects, the memory has to be rejected. But if there are a few defects, it will be more efficient and cost reducing for the company to use it by repairing. Therefore, laser-repair process is needed for such a reason and redundancy analysis is needed to establish correct target of laser-repair process. The equipment development company had provided the redundancy analysis and each development company had developed and provided separately. So, to analyze the similar type of defects, redundancy analysis time can be very different by the manufacture. The purpose of this research is to strengthen the competitive price and to apply correlation concept in business for reducing the redundancy analysis time to repair the defects

• 한국컴퓨터정보학회논문지
• /
• 제23권4호
• /
• pp.1-8
• /
• 2018

In this paper, we propose an adaptive partition cache replacement policy and evaluate the performance of our scheme using various monitoring ratios to help lifetime extension of non-volatile main memory systems without performance degradation. The proposal combines conventional LRU (Least Recently Used) replacement policy and Early Eviction Zone (E2Z), which considers a dirty bit as well as LRU bits to select a candidate block. In particular, this paper shows the performance of non-volatile memory using various monitoring ratios and determines optimized monitoring ratio and partition size of E2Z for reducing the number of writebacks using cache hit counter logic and hit predictor. In the experiment evaluation, we showed that 1:128 combination provided the best results of writebacks and runtime, in terms of performance and complexity trade-off relation, and our proposal yielded up to 42% reduction of writebacks, compared with others.

• 대한임베디드공학회논문지
• /
• 제6권1호
• /
• pp.32-40
• /
• 2011

For a NAND booting based embedded system, an application program on the NAND flash memory is downloaded to the RAM when the system is booted. In this case, the application program exists in both the RAM and the NAND flash so the RAM usage is increased. In this paper, we suggested the demand paging technique for the decreasing of the RAM usage for OS-less NAND booting based embedded systems. As a result of a benchmark test, 40~80% of the code memory usage was reduced with below 5% of execution time delay.