JCSE, vol. 6, no. 4, pp.279-279, December, 2012

DOI:

Preface to KOCSEA Special Issue

Jungwoo Ryoo, Yoohwan Kim, Yoonsuck Choe, Bongjun Ko, Taek Jin Kwon
Penn State University/ University of Nevada/ Texas A&M University/ IBM Research/ Applied Communication Sciences

Abstract: It’s our pleasure to edit this special KOrean Computer Scientists and Engineers Association in America (KOCSEA) issue of JCSE. The theme of the special issue is High Performance Computing (HPC). KOCSEA is a not-for-profit organization of Korean American computer scientists and engineers in the United States. The purpose of KOCSEA is to: “promote communication, information exchanges and cooperation among its members, and to provide opportunities for the members to make contributions to computer-related fields in the U.S. and Korea.” Since its foundation in 1983, KOCSEA has grown both quantitatively and qualitatively. For example, KOCSEA hosts its annual technical symposium that invites renowned computer scientists and engineers who are the leaders of their respective fields in both industry and academia. KOCSEA has also been organizing the Computer Science, Information Systems, Industrial Engineering, and Management Science (CSI) track of the annual Korean-American Scientists and Engineers Association (KSEA) conferences called U.S.-Korea Conference (UKC). UKC 2012 was held in Los Angeles, California this year, and KOCSEA played an indispensable role in making its CSI track a success. In fact, most of the featured articles in this special issue were solicited among the presenters of the HPC session of the UKC 2012 CSI track. One of the featured articles by Dr. Byunghoon Park from Oak Ridge National Laboratory (ORNL) provides an overview of the many fascinating HPC topics discussed during the HPC session and offers a vision for future research directions in the HPC field. Dr. Park was the HPC session chair of the UKC 2012 CSI track and is also the chair of the KOCSEA HPC Special Interest Group (SIG). The paper by J. Park et al. presents a performance study on how individual injected tasks perform depending on the type of the existing jobs in a cluster. This research problem is an important issue from an end-user’s point of view in a cloud environment where the users care more about the turnaround time for their jobs and less about the performance of the system as a whole. Four benchmarks with mixed characteristics are used as a baseline, and the same four types are injected to measure the 4×4 conditions (times 3 different load sizes). The authors recommend scheduling/job allocation guidelines based on their results. The paper by B. Kim et al. discusses methods to derive application characteristics and use them to optimize the use of HPC. Their paper provides general guidelines for performance optimization in one-node level optimization and internode scalability improvement. The results of this research offer an insight into a systematic approach in HPC applications performance engineering. The paper by S. Rizvi proposes a power efficient algorithm for improving the computing power of the next generation of wireless receivers. This paper provides an interesting perspective in HPC, which focuses on how to rea

Keyword: no keyword

Full Paper:   146 Downloads, 2768 View