skip to main content

DOE PAGESDOE PAGES

Title: A new parallelization scheme for adaptive mesh refinement

Here, we present a new method for parallelization of adaptive mesh refinement called Concurrent Structured Adaptive Mesh Refinement (CSAMR). This new method offers the lower computational cost (i.e. wall time x processor count) of subcycling in time, but with the runtime performance (i.e. smaller wall time) of evolving all levels at once using the time step of the finest level (which does more work than subcycling but has less parallelism). We demonstrate our algorithm's effectiveness using an adaptive mesh refinement code, AMSS-NCKU, and show performance on Blue Waters and other high performance clusters. For the class of problem considered in this paper, our algorithm achieves a speedup of 1.7-1.9 when the processor count for a given AMR run is doubled, consistent with our theoretical predictions.
Authors:
 [1] ;  [2] ;  [3] ;  [4]
  1. Louisiana State Univ., Baton Rouge, LA (United States). Center for Computation and Technology
  2. Chinese Academy of Sciences (CAS), Beijing (China). Inst. of Applied Mathematics and Academy of Mathematics and Systems Science
  3. Louisiana State Univ., Baton Rouge, LA (United States). Center for Computation and Technology and Dept. of Computer Science
  4. Tsinghua Univ., Beijing (China). Tsinghua National Lab. for Information, Science and Technology and Dept. of Computer, Science and Technology
Publication Date:
Grant/Contract Number:
SC0008714; 1265449; 61272087; 61073008; 60773148; 4082016; 4122039
Type:
Published Article
Journal Name:
Journal of Computational Science
Additional Journal Information:
Journal Volume: 16; Journal Issue: C; Journal ID: ISSN 1877-7503
Publisher:
Elsevier
Research Org:
Louisiana State Univ., Baton Rouge, LA (United States)
Sponsoring Org:
USDOE Office of Science (SC), Advanced Scientific Computing Research (ASCR) (SC-21); National Science Foundation (NSF); National Natural Science Foundation of China (NNSFC)
Country of Publication:
United States
Language:
English
Subject:
97 MATHEMATICS AND COMPUTING; Parallel application frameworks; Parallel algorithms; Parallel applications; Adaptive mesh refinement
OSTI Identifier:
1433730
Alternate Identifier(s):
OSTI ID: 1438051

Loffler, Frank, Cao, Zhoujian, Brandt, Steven R., and Du, Zhihui. A new parallelization scheme for adaptive mesh refinement. United States: N. p., Web. doi:10.1016/j.jocs.2016.05.003.
Loffler, Frank, Cao, Zhoujian, Brandt, Steven R., & Du, Zhihui. A new parallelization scheme for adaptive mesh refinement. United States. doi:10.1016/j.jocs.2016.05.003.
Loffler, Frank, Cao, Zhoujian, Brandt, Steven R., and Du, Zhihui. 2016. "A new parallelization scheme for adaptive mesh refinement". United States. doi:10.1016/j.jocs.2016.05.003.
@article{osti_1433730,
title = {A new parallelization scheme for adaptive mesh refinement},
author = {Loffler, Frank and Cao, Zhoujian and Brandt, Steven R. and Du, Zhihui},
abstractNote = {Here, we present a new method for parallelization of adaptive mesh refinement called Concurrent Structured Adaptive Mesh Refinement (CSAMR). This new method offers the lower computational cost (i.e. wall time x processor count) of subcycling in time, but with the runtime performance (i.e. smaller wall time) of evolving all levels at once using the time step of the finest level (which does more work than subcycling but has less parallelism). We demonstrate our algorithm's effectiveness using an adaptive mesh refinement code, AMSS-NCKU, and show performance on Blue Waters and other high performance clusters. For the class of problem considered in this paper, our algorithm achieves a speedup of 1.7-1.9 when the processor count for a given AMR run is doubled, consistent with our theoretical predictions.},
doi = {10.1016/j.jocs.2016.05.003},
journal = {Journal of Computational Science},
number = C,
volume = 16,
place = {United States},
year = {2016},
month = {5}
}