The ScaLAPACK (or Scalable LAPACK) library includes a subset of LAPACK routines redesigned for distributed memory parallel computers. It is currently written in a Single-Program-Multiple-Data style using explicit message passing for interprocessor communication. It assumes matrices are laid out in a two-dimensional block cyclic decomposition. The goal is to have ScaLAPACK routines resemble their LAPACK equivalents as much as possible. Just as LAPACK is built on top of the BLAS, ScaLAPACK relies on the PBLAS (Parallel Basic Linear Algebra Subprograms) and the BLACS (Basic Linear Algebra Communication Subprograms). The PBLAS perform computations analogous to the BLAS but on matrices distributed across multiple processors. The PBLAS rely on the communication protocols of the BLACS. The BLACS are designed for linear algebra applications and provide portable communication across a wide variety of distributed-memory architectures. At the present time, they are available for the Intel Gamma, Delta, and Paragon, Thinking Machines CM-5, IBM SPs, and PVM. They will soon be available for the CRAY T3D. For more information:
echo ''send index from scalapack'' | mail email@example.comAll questions/comments can be directed to firstname.lastname@example.org.