Some basic facts for efficient massively parallel computation

  title={Some basic facts for efficient massively parallel computation},
  author={Maya Neytcheva},
Four fundamental aspects of eecient massively parallel computation are discussed: (i) the need for massively parallel computations and, consequently, the need for distributed memory machines (ii) the need for scalable algorithms (iii) some physical limits to massively parallel computations (iv) the need for dynamic load balancing algorithms. Massively parallel computation in large scale numerical modelling requires scal-able algorithms, i.e., the performance of algorithms must increase… CONTINUE READING


Publications referenced by this paper.
Showing 1-10 of 12 references

The stabilized V-cycle method

O. Axelsson
TICAM Conference Proceedings, editor J.T. Oden, Wiley, • 1996
View 4 Excerpts
Highly Influenced

Arithmetic and communication complexity of pre conditioning methods, Ph.D. Thesis, University of Nijmegen

M. Neytcheva
View 3 Excerpts
Highly Influenced

Some basic facts for e cient mas sively parallel computation, Report 9607, University of Nijmegen

O. Axelsson, M. Neytcheva
View 2 Excerpts

A uni ed geometric ap proach to graph separators

Gary L. Miller, Stephen A. Vavasis
In 32th Annual Symposiium on Foundations of Computer Science, • 1991

Algebraic multilevel precondi tioning methods II

P.S.O. Axelsson
Vassilevski • 1990
View 1 Excerpt

Similar Papers

Loading similar papers…