TY - GEN
T1 - ForestClaw: Hybrid forest-of-octrees AMR for hyperbolic conservation laws
AU - Burstedde, Carsten
AU - Calhoun, Donna
AU - Mandli, Kyle
AU - Terrel, Andy R.
N1 - KAUST Repository Item: Exported on 2021-09-21
Acknowledgements: We would like to thank the Texas Advanced Computing Center (TACC) for access to the Stampede supercomputer under allocations TG-DPP130002 and TG-ASC130001 granted by the NSF XSEDE program. The authors acknowledge valuable discussion with Randy LeVeque, Marsha Berger, and Hans-Petter Langtangen. We also acknowledge David Ketcheson and the KAUST sponsored HPC3 numerics workshop at which the initial phases of this project were first discussed. The second author would like to also acknowledge the Isaac Newton Institute (Cambridge, UK), where much of the preliminary development work for ForestClaw was done. The fourth author recognizes Simula Research Lab, Norway, for funding. The leaf/patch paradigm was independently presented by B. as part of a talk at the SCI Institute, Salt Lake City, Utah, USA in July 2011.
This publication acknowledges KAUST support, but has no KAUST affiliated authors.
PY - 2014
Y1 - 2014
N2 - We present a new hybrid paradigm for parallel adaptive mesh refinement (AMR) that combines the scalability and lightweight architecture of tree-based AMR with the computational efficiency of patch-based solvers for hyperbolic conservation laws. The key idea is to interpret each leaf of the AMR hierarchy as one uniform compute patch in Rd with md degrees of freedom, where m is customarily between 8 and 32. Thus, computation on each patch can be optimized for speed, while we inherit the flexibility of adaptive meshes. In our work we choose to integrate with the p4est AMR library since it allows us to compose the mesh from multiple mapped octrees and enables the cubed sphere and other nontrivial multiblock geometries. We describe aspects of the parallel implementation and close with scalings for both MPI-only and OpenMP/MPI hybrid runs, where the largest MPI run executes on 16,384 CPU cores. © 2014 The authors and IOS Press.
AB - We present a new hybrid paradigm for parallel adaptive mesh refinement (AMR) that combines the scalability and lightweight architecture of tree-based AMR with the computational efficiency of patch-based solvers for hyperbolic conservation laws. The key idea is to interpret each leaf of the AMR hierarchy as one uniform compute patch in Rd with md degrees of freedom, where m is customarily between 8 and 32. Thus, computation on each patch can be optimized for speed, while we inherit the flexibility of adaptive meshes. In our work we choose to integrate with the p4est AMR library since it allows us to compose the mesh from multiple mapped octrees and enables the cubed sphere and other nontrivial multiblock geometries. We describe aspects of the parallel implementation and close with scalings for both MPI-only and OpenMP/MPI hybrid runs, where the largest MPI run executes on 16,384 CPU cores. © 2014 The authors and IOS Press.
UR - http://hdl.handle.net/10754/671340
UR - https://www.medra.org/servlet/aliasResolver?alias=iospressISSNISBN&issn=0927-5452&volume=25&spage=253
UR - http://www.scopus.com/inward/record.url?scp=84902247606&partnerID=8YFLogxK
U2 - 10.3233/978-1-61499-381-0-253
DO - 10.3233/978-1-61499-381-0-253
M3 - Conference contribution
SN - 9781614993803
SP - 253
EP - 262
BT - International Conference on Parallel Programming (ParCo)
PB - IOS Press BV
ER -