• Journal of KIISE:Computer Systems and Theory
• /
• v.37 no.2
• /
• pp.90-102
• /
• 2010

Flash memory has been widely used as an important storage device for consumer electronics. For the flash memory-based storage systems, FTL (Flash Translation Layer) is used to handle the mapping between a logical page address and a physical page address. Especially, log buffer-based FTLs provide a good performance with small-sized mapping information. In designing the log buffer-based FTL, one important factor is to determine the mapping structure between data blocks and log blocks, called associativity. While previous works use static associativity fixed at the design time, we propose a new log block mapping scheme which adjusts associativity based on the run-time workload. Our proposed scheme improves the I/O performance about 5~16% compared to the static scheme by adjusting the associativity to provide the best performance.

• The Transactions of the Korea Information Processing Society
• /
• v.3 no.1
• /
• pp.118-126
• /
• 1996

Memory hierarchy is a useful mechanism for improving the memory access speed and making the program space larger by layering the memories and separating program spaces from memory spaces. However, it needs at least two memory accesses for each data reference : a TLB(Translation Lookaside Buffer) access for the address translation and a data cache access for the desired data. If the cache size increases to the multiplication of page size and the cache associativity, it is difficult to access the TLB with the cache in parallel, thereby making longer the critical timing path in the processor. To achieve such parallel accesses, we present the hybrid mapped TLB which combines a direct mapped TLB with a very small fully-associative mapped TLB. The former can reduce the TLB access time. while the latter removes the conflict misses from the former. The trace-driven simulation shows that under given workloads the proposed TLB is effective even when a fully-associative mapped TLB with only four entries is added because the effects of its increased misses are offset by its speed benefits.

• Journal of KIISE
• /
• v.42 no.3
• /
• pp.379-389
• /
• 2015

With the advent of non-volatile memories such as Phase Change Memory (PCM or PRAM) and Magneto Resistive RAM (MRAM), active studies have been carried out on how to replace Dynamic Random-Access Memory (DRAM) with PRAM. In this paper, we study both endurance and performance issues of existing join algorithms that are based on PRAM-based computer systems and have been widely used until now: Block Nested Loop Join, Sort-Merge Join, Grace Hash Join, and Hybrid Hash Join. Our experimental results show that the existing join algorithms need to be redesigned to improve both the endurance and performance of PRAMs. To the best of our knowledge, this is the first research to scientifically study the results of the four join algorithms running on PRAM-based systems. In this work, our main contribution is the modeling and implementation of a PRAM-based simulator for a comparative study of the existing join algorithms.

• Journal of KIISE:Computing Practices and Letters
• /
• v.14 no.7
• /
• pp.636-645
• /
• 2008

OneNAND flash is a high-performance hybrid flash memory that combines the advantages of both NAND flash and NOR flash. OneNAND flash has not only all virtues of NAND flash but also greatly enhanced read performance which is considered as a downside of NAND flash. As a result, it is widely used in mobile applications such as mobile phones, digital cameras, PMP, and portable game players. However, most of the general purpose operating systems, such as Linux, can not exploit the read performance of OneNAND flash because of the restrictions imposed by their virtual memory system and block I/O architecture. In order to solve that problem, we suggest a new approach called virtual I/O segment. By using virtual I/O segment, the superior read performance of OneNAND flash can be exploited without modifying the existing block I/O architecture and MTD subsystem. Experiments by implementations show that this approach can reduce read latency of OneNAND flash as much as 54%.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.38 no.1
• /
• pp.71-79
• /
• 2014

A parallelization of the bi-conjugate gradient solver for the pressure equation of the CUPID (component unstructured program for interfacial dynamics) code, which was developed for analyzing the components of a pressurized water-cooled reactor, was studied in a symmetric multi-processing system. The parallel performance was investigated for three typical parallel programming models (MPI, OpenMP, Hybrid) by solving incompressible backward-facing step flow at various grid resolutions. It was confirmed that parallel performance was low when problem size was small or the memory requirement for each thread was considerably higher than the cache memory. Furthermore, it was shown that MPI was better than OpenMP regardless of the problem size, and Hybrid was the best when the number of threads was relatively small.

• Journal of the Korea Computer Graphics Society
• /
• v.21 no.5
• /
• pp.37-44
• /
• 2015

This paper presents a novel 3D model representation, called hybrid model representation, to overcome existing 3D volume-based indoor scene reconstruction mechanism. In indoor 3D scene reconstruction, volume-based model representation can reconstruct detailed 3D model for the narrow scene. However it cannot reconstruct large-scale indoor scene due to its memory consumption. This paper presents a memory efficient plane-hash model representation to enlarge the scalability of the indoor scene reconstruction. Also, the proposed method uses plane-hash model representation to reconstruct large, structural planar objects, and at the same time it uses volume-based model representation to recover small detailed region. Proposed method can be implemented in GPU to accelerate the computation and reconstruct the indoor scene in real-time.

• Journal of KIISE:Computing Practices and Letters
• /
• v.15 no.12
• /
• pp.918-922
• /
• 2009

Naturally, block mapping FTL works well for sequential writes while page mapping FTL does well for random writes. To exploit their advantages, a practical FTL should be able to selectively apply a suitable scheme between page and block mappings for each write pattern. To meet that requirement, we propose a hybrid mapping FTL, which we call Janus-FTL, that distributes data to either block or page mapping areas. Also, we propose the fusion operation to relocate the data from block mapping area to page mapping area and the defusion operation to relocate the data from page mapping area to block mapping area. And experimental results of Janus-FTL show performance improvement of maximum 50% than other hybrid mapping FTLs.

• Journal of KIISE
• /
• v.42 no.8
• /
• pp.947-953
• /
• 2015

PRAM is being considered as a potential successor to DRAM because of its characteristics such as byte-addressability, non-volatility, and high density. To gain its benefits, buffer cache replacement algorithm based on PRAM has been actively studied. However, most of the previous studies on buffer cache replacement algorithm limitedly exploit the byte-level performance of PRAM by focusing its limited lifetime and slower access latency compared to DRAM. In this paper, we propose a novel buffer cache replacement algorithm that fully considers the byte-level performance of PRAM and the performance of secondary storage. To take advantage of small size write on PRAM, proposed scheme keeps pages, which are frequently accessed with a small size write, on PRAM and allows the selective page migration from DRAM to PRAM. As a result, our scheme significantly reduces the number of PRAM writes. Our experimental results indicate for real workloads that our scheme reduces the number of PRAM writes by up to 92% and improves its performance by up to 62% compared to CLOCK.

• Nuclear Engineering and Technology
• /
• v.55 no.9
• /
• pp.3423-3440
• /
• 2023

In this work, a multivariate time-series machine learning meta-model is developed to predict the transient response of a typical nuclear power plant (NPP) undergoing a steam generator tube rupture (SGTR). The model employs Recurrent Neural Networks (RNNs), including the Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and a hybrid CNN-LSTM model. To address the uncertainty inherent in such predictions, a Bayesian Neural Network (BNN) was implemented. The models were trained using a database generated by the Best Estimate Plus Uncertainty (BEPU) methodology; coupling the thermal hydraulics code, RELAP5/SCDAP/MOD3.4 to the statistical tool, DAKOTA, to predict the variation in system response under various operational and phenomenological uncertainties. The RNN models successfully captures the underlying characteristics of the data with reasonable accuracy, and the BNN-LSTM approach offers an additional layer of insight into the level of uncertainty associated with the predictions. The results demonstrate that LSTM outperforms GRU, while the hybrid CNN-LSTM model is computationally the most efficient. This study aims to gain a better understanding of the capabilities and limitations of machine learning models in the context of nuclear safety. By expanding the application of ML models to more severe accident scenarios, where operators are under extreme stress and prone to errors, ML models can provide valuable support and act as expert systems to assist in decision-making while minimizing the chances of human error.

• Journal of the Korea Society of Computer and Information
• /
• v.16 no.10
• /
• pp.23-31
• /
• 2011

Recently, Solid state disk is mainly used because this device has lower power consumption as well as higher response time. But it features higher price and lower performance at delete and write operations compared with HDD. To compensate this defect, Hybrid hard disk with internal non-volatile flash memory was issued. This NVCache is used as a kind of cache for disk blocks. In this paper, an I/O scheme for H-HDD is proposed for improving low power consumption as well as response time. Our method is to use this NVCache as read cache mainly and write cache when write requests are concentrated. In read cache operation, disk blocks with higher priority determined on basis of time as well as spatial localities are prefetched, which can improve response time. The write operation is conducted only at write peak time as disk spindle up costs higher battery power as well as response time. Experiments results show that the suggested method can improve response time of H-HDD and lower the power consumption.