Excellence in Research and Innovation for Humanity

Sungho Kim

Publications

5

Publications

5
1854
Performance Analysis of the Subgroup Method for Collective I/O
Abstract:

As many scientific applications require large data processing, the importance of parallel I/O has been increasingly recognized. Collective I/O is one of the considerable features of parallel I/O and enables application programmers to easily handle their large data volume. In this paper we measured and analyzed the performance of original collective I/O and the subgroup method, the way of using collective I/O of MPI effectively. From the experimental results, we found that the subgroup method showed good performance with small data size.

Keywords:
Collective I/O, MPI, parallel file system.
4
7619
CNet Module Design of IMCS
Abstract:
IMCS is Integrated Monitoring and Control System for thermal power plant. This system consists of mainly two parts; controllers and OIS (Operator Interface System). These two parts are connected by Ethernet-based communication. The controller side of communication is managed by CNet module and OIS side is managed by data server of OIS. CNet module sends the data of controller to data server and receives commend data from data server. To minimizes or balance the load of data server, this module buffers data created by controller at every cycle and send buffered data to data server on request of data server. For multiple data server, this module manages the connection line with each data server and response for each request from multiple data server. CNet module is included in each controller of redundant system. When controller fail-over happens on redundant system, this module can provide data of controller to data sever without loss. This paper presents three main features – separation of get task, usage of ring buffer and monitoring communication status –of CNet module to carry out these functions.
Keywords:
Ethernet communication, DCS, power plant, ring buffer, data integrity
3
11678
Analysis of Long-Term File System Activities on Cluster Systems
Abstract:
I/O workload is a critical and important factor to analyze I/O pattern and to maximize file system performance. However to measure I/O workload on running distributed parallel file system is non-trivial due to collection overhead and large volume of data. In this paper, we measured and analyzed file system activities on two large-scale cluster systems which had TFlops level high performance computation resources. By comparing file system activities of 2009 with those of 2006, we analyzed the change of I/O workloads by the development of system performance and high-speed network technology.
Keywords:
I/O workload, Lustre, GPFS, Cluster File System
2
11793
Feature Point Detection by Combining Advantages of Intensity-based Approach and Edge-based Approach
Abstract:
In this paper, a novel corner detection method is presented to stably extract geometrically important corners. Intensity-based corner detectors such as the Harris corner can detect corners in noisy environments but has inaccurate corner position and misses the corners of obtuse angles. Edge-based corner detectors such as Curvature Scale Space can detect structural corners but show unstable corner detection due to incomplete edge detection in noisy environments. The proposed image-based direct curvature estimation can overcome limitations in both inaccurate structural corner detection of the Harris corner detector (intensity-based) and the unstable corner detection of Curvature Scale Space caused by incomplete edge detection. Various experimental results validate the robustness of the proposed method.
Keywords:
Feature, intensity, contour, hybrid.
1
14259
Design and Implementation of Shared Memory based Parallel File System Logging Method for High Performance Computing
Abstract:
I/O workload is a critical and important factor to analyze I/O pattern and file system performance. However tracing I/O operations on the fly distributed parallel file system is non-trivial due to collection overhead and a large volume of data. In this paper, we design and implement a parallel file system logging method for high performance computing using shared memory-based multi-layer scheme. It minimizes the overhead with reduced logging operation response time and provides efficient post-processing scheme through shared memory. Separated logging server can collect sequential logs from multiple clients in a cluster through packet communication. Implementation and evaluation result shows low overhead and high scalability of this architecture for high performance parallel logging analysis.
Keywords:
I/O workload, PVFS, I/O Trace.