• KIPS Transactions on Computer and Communication Systems
• /
• v.10 no.5
• /
• pp.123-136
• /
• 2021

RPC (Remote Procedure Call)-based Graphics Processing Unit (GPU) virtualization technology is one of the technologies for sharing GPUs with multiple user virtual machines. However, in a cloud environment, unlike CPU or memory, general GPUs do not provide a resource isolation technology that can limit the resource usage of virtual machines. In particular, in an RPC-based virtualization environment, since GPU tasks executed in each virtual machine are performed in the form of multi-process, the lack of resource isolation technology causes performance degradation due to resource competition. In addition, the GPU memory competition accelerates the performance degradation as the resource demand of the virtual machines increases, and the fairness decreases because it cannot guarantee equal performance between virtual machines. This paper, in the RPC-based GPU virtualization environment, analyzes the performance degradation problem caused by resource contention when the GPU memory requirement of virtual machines exceeds the available GPU memory capacity and proposes a GPU memory management technique to solve this problem. Also, experiments show that the GPU memory management technique proposed in this paper can improve the performance of GPGPU tasks.

• JSTS:Journal of Semiconductor Technology and Science
• /
• v.15 no.5
• /
• pp.476-484
• /
• 2015

Due to the recent advances in non-volatile memory technologies such as PCM, a new memory hierarchy of computer systems is expected to appear. In this paper, we explore the performance of PCM-based swap systems and discuss how this system can be managed efficiently. Specifically, we introduce three management techniques. First, we show that the page fault handling time can be reduced by attaching PCM on DIMM slots, thereby eliminating the software stack overhead of block I/O and the context switch time. Second, we show that it is effective to reduce the page size and turn off the read-ahead option under the PCM swap system where the page fault handling time is sufficiently small. Third, we show that the performance is not degraded even with a small DRAM memory under a PCM swap device; this leads to the reduction of DRAM's energy consumption significantly compared to HDD-based swap systems. We expect that the result of this paper will lead to the transition of the legacy swap system structure of "large memory - slow swap" to a new paradigm of "small memory - fast swap."

• Journal of the Korean Society for information Management
• /
• v.20 no.4 s.50
• /
• pp.91-111
• /
• 2003

This paper shows that digital libraries can be understood through STIN models which emphasize interactions among components in networks. The enrollment strategies in the American Memory make human and non-human factors interact. Specifically, this paper articulates that the relationships between users and collections, between users and staff, and between users and users are closely linked through the strategies . Observing the linkages among these components ,this paper found that the enrollment processes not only draw users to the American Memory, but also alter roles of components and creates new roles and players for them. The alterations of roles and the resulting changes of relationships among components mean that digital libraries lead to transform the grounding of knowledge works in a society.

• Proceedings of the KIEE Conference
• /
• 1998.11b
• /
• pp.672-674
• /
• 1998

Java virtual machine has the garbage collector that automate memory management. Mark-compact algorithm is one of the garbage collection algorithm that operating in 2 phases, marking and sweeping. One is Marking is marking live objects reachable from root object set. Sweeping is sweeping unmarked object from memory(return to free memory pool). This algorithm is easy to implement but cause a memory fragmentation. So compacting memory, before memory defragmentation become serious. When compacting memory, all other processes are suspended. It is critical for embedded system that must guarantee real-time processing. This paper introduce enhanced mark-compact garbage collection algorithm. Grouping the objects by their size that minimize memory fragmentation. Then apply smart algorithm to the grouped objects when allocating objects and compacting memory.

• Knowledge Management Research
• /
• v.16 no.2
• /
• pp.67-89
• /
• 2015

Improving productivity of knowledge workers is an important issue in the 21st century referred as knowledge-based society. The core key word is knowledge sharing among constituents of an organization. The purpose of this study is to combine the social network position factors with attitude and behavior factors, and develop an integrated research model for the knowledge sharing among members of an organization. This study adopted the integrated theoretical framework based on social capital, self-efficacy, transactive memory, and knowledge sharing. Surveys were conducted to 42 organizational members from a department in a leading IT outsourcing company to empirically test the proposed research model. In order to validate the proposed research model, social network analysis tool, UCINET, a structural equation modeling tool, SmartPLS, were utilized. The empirical result showed that, first of all, organizational members' familiarity network position had significant influence on knowledge self-efficacy and transactive memory capability. Second, knowledge self-efficacy and transactive memory capability affected knowledge sharing intention. Third, knowledge sharing intention also had an impact on the job performance. However, organizational members' expertise network position had no significant influence on knowledge self-efficacy and transactive memory capability. This finding reveals the importance of the emotional approach rather than the rational approach in knowledge management. The theoretical and practical implications on the research findings were discussed along with limitations.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.28B no.9
• /
• pp.703-711
• /
• 1991

In this paper, an efficient storage reclamation algorithm for RISC parallel processing in the object orented programming environments is presented. The memory management for the dynamic memory allocation and the frequent memory access in object oriented programming is the main factor that decreases RISC parallel processing performance. The proposed algorithm can be efficiently allocated the memory space of RISCy computer which is required the frequent memory access, so it can be increased RISC parallel processing performance. The proposed algorithm is verified the efficiency by implementing C language on SUN SPARC(4.3 BSD UNIX).

• The KIPS Transactions:PartD
• /
• v.15D no.5
• /
• pp.589-598
• /
• 2008

Recently, Mobile Computing Devices are used generally. And Information which is processed by Mobile computing devices is increasing. Because information is digitalizing. So Mobile computing Devices demand an Embedded DBMS for efficient management of information. Moreover Mobile computing Devices demand an efficient storage management in NAND-type flash memory because the NAND-type flash memory is using generally in Mobile computing devices and the NAND-type flash memory is more expensive than the magnetic disks. So that in this paper, we present an efficient Compressed Data Management System for the embedded DBMS that is used in flash memory. This proposed system improve the space utilization and extend a lifetime of a flash memory because it decreases the size of data.

• International Journal of Contents
• /
• v.12 no.1
• /
• pp.60-64
• /
• 2016

As new applications of cloud data management system (CDMS) such as online games, cooperation edit, social network, and so on, are increasing, transaction processing capabilities for CDMS are required. Several transaction processing methods for cloud data management system (CDMS) have been proposed. However, existing transaction processing methods have some problems. Some of them provide limited transaction processing capabilities. Some of them are hard to be integrated with existing CDMSs. In this paper, we proposed a new concurrency control method to support transaction processing capability for CDMS to solve these problems. The proposed method was designed and implemented based on Spark, an in-memory distributed processing framework. It uses RDD (Resilient Distributed Dataset) model to provide fault tolerant to data in the main memory. In our proposed method, database stored in CDMS is loaded to main memory managed by Spark. The loaded data set is then transformed to RDD. In addition, we proposed a multi-version concurrency control method through immutable characteristics of RDD. Finally, we performed experiments to show the feasibility of the proposed method.

• Proceedings of the KIEE Conference
• /
• 2007.04a
• /
• pp.157-159
• /
• 2007

The conventional hard disk has been the dominant database storage system for over 25 years. Recently, hybrid systems which incorporate the advantages of flash memory into the conventional hard disks are considered to be the next dominant storage systems to support databases for desktops and server computers. Their features are satisfying the requirements like enhanced data I/O, energy consumption and reduced boot time, and they are sufficient to hybrid storage systems as major database storages. However, we need to improve traditional index node management schemes based on B-Tree due to the relatively slow characteristics of hard disk operations, as compared to flash memory. In order to achieve this goal, we propose a new index node management scheme called FNC-Tree. FNC-Tree-based index node management enhanced search and update performance by caching data objects in unused free area of flash leaf nodes to reduce slow hard disk I/Os in index access processes.

• ETRI Journal
• /
• v.33 no.2
• /
• pp.230-239
• /
• 2011

Dynamic memory allocators for real-time embedded systems need to fulfill three fundamental requirements: bounded worst-case execution time, fast average execution time, and minimal fragmentation. Since embedded systems generally run continuously during their whole lifetime, fragmentation is one of the most important factors in designing the memory allocator. This paper focuses on minimizing fragmentation while other requirements are still satisfied. To minimize fragmentation, a part of a memory region is segregated by the proposed budgeting method that exploits the memory profile of the given application. The budgeting method can be applied for any existing memory allocators. Experimental results show that the memory efficiency of allocators can be improved by up to 18.85% by using the budgeting method. Its worst-case execution time is analyzed to be bounded.