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

Title: Predicting the Performance Impact of Different Fat-Tree Configurations

Authors:
; ; ; ; ; ; ; ; ;
Publication Date:
Research Org.:
Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States)
Sponsoring Org.:
USDOE
OSTI Identifier:
1410072
Report Number(s):
LLNL-CONF-736289
DOE Contract Number:
AC52-07NA27344
Resource Type:
Conference
Resource Relation:
Conference: Presented at: Supercomputing, Denver, CO, United States, Nov 12 - Nov 17, 2017
Country of Publication:
United States
Language:
English
Subject:
97 MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE

Citation Formats

Jain, N, Bhatele, A, Howell, L, Boehme, D, Leon, E, Karlin, I, Mubarak, M, Wolfe, N, Gamblin, T, and Leininger, M. Predicting the Performance Impact of Different Fat-Tree Configurations. United States: N. p., 2017. Web. doi:10.1145/3126908.3126967.
Jain, N, Bhatele, A, Howell, L, Boehme, D, Leon, E, Karlin, I, Mubarak, M, Wolfe, N, Gamblin, T, & Leininger, M. Predicting the Performance Impact of Different Fat-Tree Configurations. United States. doi:10.1145/3126908.3126967.
Jain, N, Bhatele, A, Howell, L, Boehme, D, Leon, E, Karlin, I, Mubarak, M, Wolfe, N, Gamblin, T, and Leininger, M. Thu . "Predicting the Performance Impact of Different Fat-Tree Configurations". United States. doi:10.1145/3126908.3126967. https://www.osti.gov/servlets/purl/1410072.
@article{osti_1410072,
title = {Predicting the Performance Impact of Different Fat-Tree Configurations},
author = {Jain, N and Bhatele, A and Howell, L and Boehme, D and Leon, E and Karlin, I and Mubarak, M and Wolfe, N and Gamblin, T and Leininger, M},
abstractNote = {},
doi = {10.1145/3126908.3126967},
journal = {},
number = ,
volume = ,
place = {United States},
year = {Thu Mar 30 00:00:00 EDT 2017},
month = {Thu Mar 30 00:00:00 EDT 2017}
}

Conference:
Other availability
Please see Document Availability for additional information on obtaining the full-text document. Library patrons may search WorldCat to identify libraries that hold this conference proceeding.

Save / Share:
  • The authors present an algorithm for the construction of a 3D bronchial tree representation based on an asymmetric dichotomy model proposed by Weibel. The model has a total of ten branch generations and three independent parameters for each branch. The bronchial tree model is constructed by dividing each branch of a generation into two subbranches, except for the first few generations where some branches divided into three subbranches. The bronchial tree model is used as a testground to simulate different CT acquisition/reconstruction schemes and examine the role of CT parameters, such as slice thickness and overlap/spacing, on the measurement ofmore » fractal properties of the lung. A digital version of the model with high resolution (0.2 mm/pix) is created first. Acquisition schemes are then simulated by dividing this high resolution version of the model into cross-section and summing one or more cross-sections to form a series of 2D slices. Three acquisition schemes are simulated by choosing different ways to generate these 2D slices.« less
  • In order to compare different PET camera designs (ring and planar geometry), this paper proposes ''figures of merit which allow questions such as ''Is it better, given an particular design, to achieve a coincidence rate of 10 kHz with 5% randoms, or a 20 kHz rate with 10% randoms.'' to be answered. The authors propose a methodology based on information theory. The image is the three dimensional distribution which conveys the information to the human or electronic observer. The image is supposed to consist of discrete image elements (voxels) of uniform size, and each characterized by a coincidence density. Themore » performance of a non-ideal imager can be studied by evaluating the effect on the visibility surface: the number of events per image element will be affected by the limited efficiency, geometrical acceptance angle, temporal resolution, energy resolution. The authors show that the effect of the degradations can be introduced in the formula by replacing the number of true coincidences by an effective number of coincidences. The non-ideal imager will distort the visibility surface, so the authors compare the performance of different PET cameras by comparing the different distortions which they induce and hence their ability to detect the information present in study objects.« less
  • The power and procurement cost of bandwidth in system-wide networks has forced a steady drop in the byte/flop ratio. This trend of computation becoming faster relative to the network is expected to hold. In this paper, we explore how cost-oriented task placement enables reducing the cost of system-wide networks by enabling high performance even on tapered topologies where more bandwidth is provisioned at lower levels. We describe APHiD, an efficient hierarchical placement algorithm that uses new techniques to improve the quality of heuristic solutions and reduces the demand on high-level, expensive bandwidth in hierarchical topologies. We apply APHiD to amore » tapered fat-tree, demonstrating that APHiD maintains application scalability even for severely tapered network configurations. Using simulation, we show that for tapered networks APHiD improves performance by more than 50% over random placement and even 15% in some cases over costlier, state-of-the-art placement algorithms.« less