Optimizing the Black-box FMM for Smooth and Oscillatory Kernels

Abstract : A black-box FMM for smooth kernels was introduced in [Fong and Darve, 2009] and has been extended to oscillatory kernels in [Messner, Schanz and Darve, 2012]. Its major advantages are the easy implementation and adap- tation for new kernels. However, it requires significantly more floating point operations compared to other FMMs. In my talk, I will present ways to tackle this drawback: (1) The exploitation of symmetries allow us to reduce the pre-computation time by a factor greater than 1000. (2) Blocking schemes increase the applicability of optimized Level 3 BLAS routines and lead to a great performance of the actual matrix-vector product.
Complete list of metadatas

https://hal.inria.fr/hal-00799885
Contributor : Olivier Coulaud <>
Submitted on : Tuesday, March 12, 2013 - 5:38:45 PM
Last modification on : Thursday, March 22, 2018 - 3:43:00 PM

Identifiers

  • HAL Id : hal-00799885, version 1

Collections

Citation

Eric Darve, Matthias Messner, Martin Schanz, Olivier Coulaud. Optimizing the Black-box FMM for Smooth and Oscillatory Kernels. SIAM Conference on Computational Science and Engineering (SIAM CSE 2013), Feb 2013, Boston, United States. ⟨hal-00799885⟩

Share

Metrics

Record views

576