Accelerating the continuous data in a SCI-based multimedia system

Sung Woo Jung, H. S. Kim, C. S. Jhon

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

The SCI(Scalable Coherent Interface) is widely used in high performance clustered systems these days. However, there has not been any consideration for scheduling policies of improving transmission performance of continuous data in a SCI-based multimedia server. The scheduling policy that does not take into account the property of continuous data may hurt the overall performance. In this paper, we propose two scheduling policies. The first policy is to maintain separate queues in a node in order to provide differentiated transmission services to continuous data and discrete data. The second one is to let continuous data even preempt discrete data that is being transmitted, which may bring more timely service to continuous data. We measured how the continuous data latency is reduced by simulations, which show that 8∼70% reduction is expected, on an average, with the first scheduling policy and observed even better 22∼99% reduction with the second scheduling policy.

Original languageEnglish
Title of host publicationProceedings of the Internatoinal Conference on Parallel and Distributed Systems - ICPADS
Pages171-176
Number of pages6
Publication statusPublished - 2001 Jan 1
Externally publishedYes
Event8th International Conference on Parallel and Distributed Systems - Kyongju, Korea, Republic of
Duration: 2001 Jun 262001 Jun 29

Other

Other8th International Conference on Parallel and Distributed Systems
CountryKorea, Republic of
CityKyongju
Period01/6/2601/6/29

ASJC Scopus subject areas

  • Engineering(all)

Fingerprint Dive into the research topics of 'Accelerating the continuous data in a SCI-based multimedia system'. Together they form a unique fingerprint.

  • Cite this

    Jung, S. W., Kim, H. S., & Jhon, C. S. (2001). Accelerating the continuous data in a SCI-based multimedia system. In Proceedings of the Internatoinal Conference on Parallel and Distributed Systems - ICPADS (pp. 171-176)