Document Detail


Streamline integration using MPI-hybrid parallelism on a large multicore architecture.
MedLine Citation:
PMID:  21885895     Owner:  NLM     Status:  PubMed-not-MEDLINE    
Abstract/OtherAbstract:
Streamline computation in a very large vector field data set represents a significant challenge due to the nonlocal and data-dependent nature of streamline integration. In this paper, we conduct a study of the performance characteristics of hybrid parallel programming and execution as applied to streamline integration on a large, multicore platform. With multicore processors now prevalent in clusters and supercomputers, there is a need to understand the impact of these hybrid systems in order to make the best implementation choice. We use two MPI-based distribution approaches based on established parallelization paradigms, parallelize over seeds and parallelize over blocks, and present a novel MPI-hybrid algorithm for each approach to compute streamlines. Our findings indicate that the work sharing between cores in the proposed MPI-hybrid parallel implementation results in much improved performance and consumes less communication and I/O bandwidth than a traditional, nonhybrid distributed implementation.
Authors:
David Camp; Christoph Garth; Hank Childs; Dave Pugmire; Kenneth I Joy
Related Documents :
24959825 - The subtalar joint axis palpation technique: part 2 - results on reliability and validi...
24919205 - Improving the recognition of eating gestures using inter-gesture sequential dependencies.
10365865 - Animal models of psychopathology: the establishment, maintenance, attenuation, and pers...
12151225 - Ion channels: does each subunit do something on its own?
22094245 - Biomimetic tissues on a chip for drug discovery.
24872455 - Evaluating the utility of dynamical downscaling in agricultural impacts projections.
Publication Detail:
Type:  Journal Article; Research Support, U.S. Gov't, Non-P.H.S.    
Journal Detail:
Title:  IEEE transactions on visualization and computer graphics     Volume:  17     ISSN:  1941-0506     ISO Abbreviation:  IEEE Trans Vis Comput Graph     Publication Date:  2011 Nov 
Date Detail:
Created Date:  2011-09-02     Completed Date:  2012-01-17     Revised Date:  2012-04-24    
Medline Journal Info:
Nlm Unique ID:  9891704     Medline TA:  IEEE Trans Vis Comput Graph     Country:  United States    
Other Details:
Languages:  eng     Pagination:  1702-13     Citation Subset:  -    
Affiliation:
Lawrence Berkeley National Laboratory, Berkeley and University of California, Davis, Davis, CA 95616, USA. dcamp@lbl.gov
Export Citation:
APA/MLA Format     Download EndNote     Download BibTex
MeSH Terms
Descriptor/Qualifier:

From MEDLINE®/PubMed®, a database of the U.S. National Library of Medicine


Previous Document:  Validity and Reliability of the Omron HJ-303 Tri-Axial Accelerometer-Based Pedometer.
Next Document:  Bone structure and density via HR-pQCT in 60d bed-rest, 2-years recovery with and without countermea...