skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Performance of the Distributed Central Analysis in BaBar.

Abstract

No abstract prepared.

Authors:
; ; ; ; ; ; ; ; ; ; ; ; ;
Publication Date:
Research Org.:
Stanford Linear Accelerator Center (SLAC)
Sponsoring Org.:
USDOE
OSTI Identifier:
899573
Report Number(s):
SLAC-REPRINT-2006-210
TRN: US0702099
DOE Contract Number:
AC02-76SF00515
Resource Type:
Journal Article
Resource Relation:
Journal Name: IEEE Trans.Nucl.Sci.53:2876-2880,2006
Country of Publication:
United States
Language:
English
Subject:
43 PARTICLE ACCELERATORS; 46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND TECHNOLOGY; PERFORMANCE; DATA ANALYSIS; SI SEMICONDUCTOR DETECTORS; ACCELERATOR FACILITIES; Computing, Instrumentation,COMP, INST

Citation Formats

Khan, A., /Brunel U., Mommsen, R.K., /UC, Irvine, Gradl, W., /Edinburgh U., Fritsch, M., /Ruhr U., Bochum, Petzold, A., /Tech. U., Dresden, Roethel, W., /UC, Irvine, Smith, D.A., and /SLAC. Performance of the Distributed Central Analysis in BaBar.. United States: N. p., 2007. Web.
Khan, A., /Brunel U., Mommsen, R.K., /UC, Irvine, Gradl, W., /Edinburgh U., Fritsch, M., /Ruhr U., Bochum, Petzold, A., /Tech. U., Dresden, Roethel, W., /UC, Irvine, Smith, D.A., & /SLAC. Performance of the Distributed Central Analysis in BaBar.. United States.
Khan, A., /Brunel U., Mommsen, R.K., /UC, Irvine, Gradl, W., /Edinburgh U., Fritsch, M., /Ruhr U., Bochum, Petzold, A., /Tech. U., Dresden, Roethel, W., /UC, Irvine, Smith, D.A., and /SLAC. Wed . "Performance of the Distributed Central Analysis in BaBar.". United States. doi:.
@article{osti_899573,
title = {Performance of the Distributed Central Analysis in BaBar.},
author = {Khan, A. and /Brunel U. and Mommsen, R.K. and /UC, Irvine and Gradl, W. and /Edinburgh U. and Fritsch, M. and /Ruhr U., Bochum and Petzold, A. and /Tech. U., Dresden and Roethel, W. and /UC, Irvine and Smith, D.A. and /SLAC},
abstractNote = {No abstract prepared.},
doi = {},
journal = {IEEE Trans.Nucl.Sci.53:2876-2880,2006},
number = ,
volume = ,
place = {United States},
year = {Wed Feb 14 00:00:00 EST 2007},
month = {Wed Feb 14 00:00:00 EST 2007}
}
  • Numerous performance models have been proposed for locking algorithms in centralized database systems, but few have been developed for distributed ones. Existing results on distributed locking usually ignore the deadlock problem so as to simplify the analysis. In this paper, a new performance model for static locking in distributed database systems is developed.A queuing model is used to approximate static locking in distributed database systems without deadlocks. Then a random graph model is proposed to find the deadlock probability of each transaction. The above two models are integrated, so that given the transaction arrival rate, the response time and themore » effective throughput can be calculated.« less
  • In this article, sparse matrix vector multiply (SpMVM) is an important kernel that frequently arises in high performance computing applications. Due to its low arithmetic intensity, several approaches have been proposed in literature to improve its scalability and efficiency in large scale computations. In this paper, our target systems are high end multi-core architectures and we use messaging passing interface + open multiprocessing hybrid programming model for parallelism. We analyze the performance of recently proposed implementation of the distributed symmetric SpMVM, originally developed for large sparse symmetric matrices arising in ab initio nuclear structure calculations. We also study important featuresmore » of this implementation and compare with previously reported implementations that do not exploit underlying symmetry. Our SpMVM implementations leverage the hybrid paradigm to efficiently overlap expensive communications with computations. Our main comparison criterion is the "CPU core hours" metric, which is the main measure of resource usage on supercomputers. We analyze the effects of topology-aware mapping heuristic using simplified network load model. Furthermore, we have tested the different SpMVM implementations on two large clusters with 3D Torus and Dragonfly topology. Our results show that the distributed SpMVM implementation that exploits matrix symmetry and hides communication yields the best value for the "CPU core hours" metric and significantly reduces data movement overheads.« less