Name
Abstract | ||
|---|---|---|
Hierarchical treecodes have, to a large extent, con- verted the compute-bound N-body problem into a memory-bound problem. The large ratio of DRAM to disk pricing suggests use of out-of-core tech- niques to overcome memory capacity limitations. We will describe a parallel, out-of-core treecode li- brary, targeted at machines with independent sec- ondary storage associated with each processor. Bor- rowing the space-filling curve techniques from our in-core library, and "manually" paging, results in ex- cellent spatial and temporal locality and very good performance. 1 Motivation N-body methods are used in the numerical simula- tion of systems ranging from the atomic to the cos- mological. In addition, the mathematical techniques developed in conjunction with the N-body problem have found application in areas as diverse as elec- tromagnetic scattering and stochastic process gener- ation. The papers collected in this mini-symposium (4), and its predecessor (1) offer ample evidence of the breadth and importance of N-body methods. A family of methods, collectively called "treecodes", use tree data structures to reduce the time required to approximately evaluate a set of in- teractions of the form: |
Year | Venue | Keywords |
|---|---|---|
1997 | PPSC | data transmission,memory management,data structure,n body problem,stochastic process,parallel processing |
Field | DocType | Citations |
Dram,Locality of reference,Data transmission,Computer science,N-body simulation,Parallel computing,Out-of-core algorithm,Memory management,Paging,Auxiliary memory | Conference | 25 |
PageRank | References | Authors |
2.93 | 0 | 2 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| John K. Salmon | 1 | 410 | 46.45 |
| Michael S. Warren | 2 | 97 | 17.16 |