HP MLIB for PA-RISC and Itanium HP-UX Version 9.5.6 Release Note

HP MLIB for PA-RISC and Itanium HP-UX Version 9.5.6 Release Note
What’s in This Version
14
subroutines for graph partitioning, mesh partitioning, and sparse matrix reordering, as
well as auxiliary routines. HP MLIB contains the same full METIS functionality as that
in the public domain METIS, however, the routine names are different. HP MLIB METIS
routine names have been prepended with mlib_ to avoid name conflict on applications
and libraries that contain their own local version of METIS.
For more information about METIS, please refer to:
http://www-users.cs.umn.edu/~karypis/metis/metis/index.html
VMATH
VMATH is a library of vector math routines corresponding to many of the widely used scalar
math routines available with C, C++, and Fortran90.
VMATH is intended for computationally intensive mathematical applications amenable to a
vector programming style.
VMATH provides two libraries: VMATH, whose interface uses 4-byte integers; and VMATH8,
whose interface uses 8-byte integers and is otherwise equivalent to VMATH. VMATH
routines come with both Fortran and C interfaces.
For more detailed information on VMATH as well as subprogram specifications, please refer
to the HP MLIB Users Guide 7th edition. Or refer to the man pages installed in the directory
/opt/mlib/share/man. VMATH man pages provide a man page for each subprogram.
SMP Parallelism
Parallel processing is available on multi-processor HP platforms. These systems can divide a
single computational process into smaller streams of execution. The result is that you can
have more than one processor executing on behalf of the same process.
Some BLAS-2, BLAS-3, LAPACK, Sparse Solvers, convolution, and FFT routines use SMP
parallelism that gives shared-memory parallel programmers a simple and flexible interface
for developing parallel applications.
For usage information, see “SMP Parallelism Usage” on page 28.
MPI Parallelism
Internal parallelism in ScaLAPACK and SuperLU_DIST is implemented using MPI—a
portable, scalable programming model that gives distributed-memory parallel programmers a
simple and flexible interface for developing parallel applications. The MPI parallel libraries
ScaLAPACK and SuperLU_DIST in HP MLIB were built and tested with HP-MPI Version
2.1.1. For MPI parallelism issues, refer to the HP-MPI documents available at
http://docs.hp.com.