IBM RS/6000 SP, Руководство пользователя

Страница: 298 / 480

274
RS/6000 SP and Clustered IBM
^
pSeries Systems Handbook
9.4.1 Operations
ESSL’s mathematical functions:
򐂰 Basic Linear Algebra Subroutines (BLAS)
򐂰 Linear Algebraic Equations
򐂰 Eigensystem Analysis
򐂰
Fourier Transforms
The ESSL products are compatible with public domain subroutine libraries, such
as Basic Linear Algebra Subprograms (BLAS), Scalable Linear Algebra Package
(ScaLAPACK), and Parallel Basic Linear Algebra Subprograms (PBLAS), Basic
Linear Algebra Communication subprograms (BLACS), and MPL/MPI making it
easy to migrate applications that utilize these libraries to use ESSL and/or
Parallel ESSL.
OSL capabilities include:
򐂰 Linear programming
򐂰
Network programming
򐂰
Mixed integer programming
򐂰 Quadratic programming
Benchmarks have shown that Parallel OSL achieves significant speed increases
on LP and MIP problems with its parallelized solution algorithms. LPs solved with
the interior point algorithm are typically able to achieve a sub-linear speed
increases of 40 to 50 percent processor utilization. That is, running such a
problem on an SP with eight processors is three times as fast as running
OSL/6000 on a single node.
Examples of applications that use these types of mathematical subroutines are:
򐂰
Structural analysis
򐂰 Time series analysis
򐂰 Computational chemistry
򐂰 Computational techniques
򐂰 Fluid dynamics analysis
򐂰
Mathematical analysis
򐂰
Seismic analysis
򐂰 Dynamic systems simulation
򐂰 Reservoir modeling