Multi-Core Software

Intel® Performance Libraries: Multi-Core-Ready Software for Numeric-Intensive Computation

GLOSSARY

BLACS: Basic Linear Algebra Communication Subprograms – a set of functions developed for ScaLAPACK which isolate the communications used by the software from the communication layer such as MPI. Used throughout MKL cluster software.

BLAS: Basic Linear Algebra Subprograms – a set of dense vector, vector matrix and matrix math functions useful in creating higher level functions such as solvers.

CODEC: COder/DECoder – used for encoding or decoding digital data streams such as video or audio.

DSS: Direct Sparse Solver – solves a system of equations in an a priori known number of operations in contrast to iterative sparse solvers for which the number of operations is data dependent.

FFT: Fast Fourier Transform – algorithms to convert, for instance, a time series into a frequency series in an efficient way.

FFTW: Fastest FFT in the West – a publicly available software package to create highly optimal FFTW. See http://www.fftw.org.

IMSL – A large commercial math software package. See http://www.vni.com

LAPACK: Linear Algebra PACKage – a set of solvers for systems of equations, eigensolvers, etc, using blocked algorithms that make effective use of the Level 3 BLAS.

LINPACK – Predates LAPACK and based on vector operations. Also a benchmark solving systems of linear equations.

METIS – A set of programs for partitioning unstructured graphs. See http://glaros.dtc.umn.edu/gkhome/views/metis

MPI: Message Passing Interface – A widely used distributed memory (cluster) communication package.

NAGLIB – A large math software package similar to IMSL. See http://www.nag.com

NETLIB – A repository of software packages such as BLAS, BLACS, LAPACK, LINPACK, ScaLAPACK and others. See http://www.netlib.org

PARDISO: PARallel DIrect SOlver – A parallel direct solver from University of Basel and licensed by MKL. See http://www.pardiso-project.org

PDE – Partial Differential Equation.

ScaLAPACK: Scalable LAPACK – cluster versions of much of LAPACK.

SIMD: Single Instruction Multiple Data – hardware to perform multiple arithmetic operations simultaneously on a single instruction, such as the SSE2 and SSE3 instructions.

Back to Top