Pipelining the Fast Multipole Method over a Runtime System

Abstract : Fast multipole methods (FMM) usually require a careful tuning of the algorithm for both the targeted physics and the hardware. In this talk, we propose a new approach that achieves high performance across architectures. We express the FMM algorithm as a task flow and employ a runtime system to process the tasks on the different pro-cessing units. We carefully design the task flow, the mathematical operators, their CPU and GPU implementations and their schedule.
Complete list of metadatas

Cited literature [1 references]  Display  Hide  Download

https://hal.inria.fr/hal-00797403
Contributor : Olivier Coulaud <>
Submitted on : Wednesday, March 6, 2013 - 2:47:55 PM
Last modification on : Thursday, January 11, 2018 - 6:22:35 AM
Long-term archiving on: Sunday, April 2, 2017 - 9:43:12 AM

File

cse.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-00797403, version 1

Citation

Emmanuel Agullo, Bérenger Bramas, Olivier Coulaud, Eric Darve, Matthias Messner, et al.. Pipelining the Fast Multipole Method over a Runtime System. SIAM Conference on Computational Science and Engineering (SIAM CSE 2013), Feb 2013, Boston, United States. ⟨hal-00797403⟩

Share

Metrics

Record views

599

Files downloads

271