• 반도체디스플레이기술학회지
• /
• 제22권3호
• /
• pp.142-148
• /
• 2023

In this paper, we identify performance issues in executing compute kernels from PolyBench, which includes compute kernels that are the core computational units of various data-intensive workloads, such as deep learning and data-intensive applications, on Processing-in-Memory (PIM) devices. Therefore, using our in-house simulator, we measured and compared the various performance metrics of workloads based on traditional out-of-order and in-order processors with Processing-in-Memory-based systems. As a result, the PIM-based system improves performance compared to other computing models due to the short-term data reuse characteristic of computational kernels from PolyBench. However, some kernels perform poorly in PIM-based systems without a multi-layer cache hierarchy due to some kernel's long-term data reuse characteristics. Hence, our evaluation and analysis results suggest that further research should consider dynamic and workload pattern adaptive approaches to overcome performance degradation from computational kernels with long-term data reuse characteristics and hidden data locality.

• 대한전자공학회논문지
• /
• 제27권6호
• /
• pp.963-971
• /
• 1990

This paper proposes a memory intensive VLSI architecture for the realization of real-time 3x3 neighborhood processor based on the distributed arithmetic. The proposed architecture is characterized by a bit serial and multi-kernel parallel processing which exploits the pixel kernel parallelism and concurrency. The chip implements 8 neighborhood processing elements in parallel with efficirnt input and output modules which operate concurrently. Besides the a4chitectural design of a neighborhood processor, the design methodology using module generator concept has been considered and MOGOT(MOdule Generator Oriented VLSI design Tool) has been constructed based on the workstation. Based on these design environments MOGOT, it has been shown that the main part of the suggested architecture can be designed efficiently using 2\ulcorner double metal CMOS technology. It includes design of input delay and data conversion module, look-up table for inner product operation, carry save accumulator, output data converter and delay module, and control module.

• Journal of Information Processing Systems
• /
• 제17권5호
• /
• pp.905-917
• /
• 2021

In recent years, edge computing technology consists of several Internet of Things (IoT) devices with embedded sensors that have improved significantly for monitoring, detection, and management in an environment where big data is commercialized. The main focus of edge computing is data optimization or task offloading due to data and task-intensive application development. However, existing offloading approaches do not consider correlations and associations between data and tasks involving edge computing. The extent of collaborative offloading segmented without considering the interaction between data and task can lead to data loss and delays when moving from edge to edge. This article proposes a range segmentation of dynamic offloading (RSDO) algorithm that isolates the offload range and collaborative edge node around the edge node function to address the offloading issue.The RSDO algorithm groups highly correlated data and tasks according to the cause of the overload and dynamically distributes offloading ranges according to the state of cooperating nodes. The segmentation improves the overall performance of edge nodes, balances edge computing, and solves data loss and average latency.

• Journal of Information Processing Systems
• /
• 제10권4호
• /
• pp.491-502
• /
• 2014

In this paper we present some research results on computing intensive applications using modern high performance architectures and from the perspective of high computational needs. Computing intensive applications are an important family of applications in distributed computing domain. They have been object of study using different distributed computing paradigms and infrastructures. Such applications distinguish for their demanding needs for CPU computing, independently of the amount of data associated with the problem instance. Among computing intensive applications, there are applications based on simulations, aiming to maximize system resources for processing large computations for simulation. In this research work, we consider an application that simulates scheduling and resource allocation in a Grid computing system using Genetic Algorithms. In such application, a rather large number of simulations is needed to extract meaningful statistical results about the behavior of the simulation results. We study the performance of Oracle Grid Engine for such application running in a Cluster of high computing capacities. Several scenarios were generated to measure the response time and queuing time under different workloads and number of nodes in the cluster.

• International Journal of Contents
• /
• 제10권2호
• /
• pp.47-58
• /
• 2014

Physiological signals provide important clues in the diagnosis and prediction of disease. Analyzing these signals is important in health and medicine. In particular, data preprocessing for physiological signal analysis is a vital issue because missing values, noise, and outliers may degrade the analysis performance. In this paper, we propose PhysioCover, a system that can recover missing values of physiological signals that were monitored in real time. PhysioCover integrates a gradual method and EM-based Principle Component Analysis (PCA). This approach can (1) more readily recover long- and short-term missing data than existing methods, such as traditional EM-based PCA, linear interpolation, 5-average and Missing Value Singular Value Decomposition (MSVD), (2) more effectively detect hidden variables than PCA and Independent component analysis (ICA), and (3) offer fast computation time through real-time processing. Experimental results with the physiological data of an intensive care unit show that the proposed method assigns more accurate missing values than previous methods.

• 정보보호학회논문지
• /
• 제18권5호
• /
• pp.181-188
• /
• 2008

E-Science Grid is designed to cope with computation-intensive tasks and to manage a huge volume of science data efficiently. However, certain tasks may involve more than one grid can offer in computation capability or incur a long wait time on other tasks. Resource sharing among Grids can solve this problem with proper processing-integrity check via audit. Due to their computing-intensive nature, the processing time of e-Science tasks tends to be long. This potential long wait before an audit failure encourages earlier audit mechanism during execution in order both to prevent resource waste and to detect any problem fast. In this paper, we propose a Light-weight, Adaptive and Reliable Audit, LARA, of processing Integrity for e-Science applications. With the LARA scheme. researchers can verify their processing earlier and fast.

• 전자통신동향분석
• /
• 제38권5호
• /
• pp.23-33
• /
• 2023

With the widespread demand from data-intensive tasks such as machine learning and large-scale databases, the amount of data processed in modern computing systems is increasing exponentially. Such data-intensive tasks require large amounts of memory to rapidly process and analyze massive data. However, existing computing system architectures face challenges when building large-scale memory owing to various structural issues such as CPU specifications. Moreover, large-scale memory may cause problems including memory overprovisioning. The Compute Express Link (CXL) allows computing nodes to use large amounts of memory while mitigating related problems. Hence, CXL is attracting great attention in industry and academia. We describe the overarching concepts underlying CXL and explore recent research trends in this technology.

• Journal of Information Processing Systems
• /
• 제19권1호
• /
• pp.1-16
• /
• 2023

Synthetic data generation is generally used in performance evaluation and function tests in data-intensive applications, as well as in various areas of data analytics, such as privacy-preserving data publishing (PPDP) and statistical disclosure limit/control. A significant amount of research has been conducted on tools and languages for data generation. However, existing tools and languages have been developed for specific purposes and are unsuitable for other domains. In this article, we propose a regular expression-based data generation language (DGL) for flexible big data generation. To achieve a general-purpose and powerful DGL, we enhanced the standard regular expressions to support the data domain, type/format inference, sequence and random generation, probability distributions, and resource reference. To efficiently implement the proposed language, we propose caching techniques for both the intermediate and database queries. We evaluated the proposed improvement experimentally.

• Journal of Information Processing Systems
• /
• 제12권1호
• /
• pp.149-168
• /
• 2016

The Active Appearance Model (AAM) is a class of deformable models, which, in the segmentation process, integrates the priori knowledge on the shape and the texture and deformation of the structures studied. This model in its sequential form is computationally intensive and operates on large data sets. This paper presents another framework to implement the standard version of the AAM model. We suggest a distributed and parallel approach justified by the characteristics of the model and their potentialities. We introduce a schema for the representation of the overall model and we study of operations that can be parallelized. This approach is intended to exploit the benefits build in the area of advanced image processing.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2012년도 춘계학술발표대회
• /
• pp.607-608
• /
• 2012

Data aggregation is a fundamental problem in wireless sensor networks which attracts great attention in recent years. Delay and energy efficiencies are two crucial issues of designing a data aggregation scheme. In this paper, we propose a distributed, energy efficient algorithm for collecting data from all sensor nodes with the minimum latency called Delay-aware Power-efficient Data Aggregation algorithm (DPDA). The DPDA algorithm minimizes the latency in data collection process by building a time efficient data aggregation network structure. It also saves sensor energy by decreasing node transmission distances. Energy is also well-balanced between sensors to achieve acceptable network lifetime. From intensive experiments, the DPDA scheme could significantly decrease the data collection latency and obtain reasonable network lifetime compared with other approaches.