Page tree
Skip to end of metadata
Go to start of metadata

ScaLAPACK (Scalable LAPACK) is a linear algebra library for parallel computers. Routines are available to diagonalize or solve dense and narrow band systems of linear equations.

ScaLAPACK implements block-oriented LAPACK linear algebra routines, adding a special set of communication routines to copy blocks of data between processors as required. Similar to LAPACK, a single subroutine call typically carries out the necessary computations.

ScaLAPACK installations include the following libraries: scalapack, redist, pblas, tools, blacs, blacsCinit, and blacsF77init.

An meeting with Professor Choi J.

I finally met Prof. Choi J. which a member of ScaLAPACK team on the 23th of June, 2008 and, He agreed to teach me some advanced topics in mathematics. I have been hearing about him for quite a long time.

  • Written in Fortran77 with a few in C
  • Covers dense and band matrices, not general sparse matrices
  • Written in a Single-Program-Multiple-Data style using explicit message passing for inter-processor communication.
  • Assumes matrices are laid out in a two-dimensional block cyclic decomposition.

Lastly, I heard that the Hama is worthwhile to continue the research on the basis of the ScaLAPACK’s ideas.

Some Performance Report

  • Performance data for Version 1.4 of ScaLAPACK on four distributed memory computers and two networks of workstations


Note that the following result is based on an internal, rudimentary experiment.



sla.solve (t=32)


real 9m2.438s, user 9m0.473s, sys 0m1.649s

real 1m55.863s, user 0m10.638s, sys 0m2.621s





Doesn't Work

Doesn't Work

  • No labels