HP MLIB for PA-RISC and Itanium2 HP-UX Version 9.5 Release Note

HP MLIB for PA-RISC and Itanium® 2 HP-UX Version 9.5 Release Note

Compatibility Information

• If your program uses subprograms from VECLIB, LAPACK, SOLVERS, VMATH,

ScaLAPACK, and SuperLU_DIST, specify -lveclib, -llapack, -lsolvers, -lvmath

-lscalapack, and -lsuperlu_dist or -lveclib8, -lveclibsc8, -llapack8,

-llapacksc8, -lsolvers8, -lvmath8, -lscalapack8, and -lsuperlu_dist8, on your

command line. For example, on an Itanium® 2-based operating system with 32- or 64-bit

addressing, link with:

% mpif90 [options] file ...-Wl,-aarchive_shared,-L/opt/mlib/lib/

-lsuperlu_dist]

NOTE DO NOT mix subprograms from the two types of 64-bit address libraries

(those with 32-bit integers and those with 64-bit integers) in the same

program.

SMP Parallelism Usage

You can enable or disable parallel processing at runtime. A program does not use parallelism

in MLIB unless parallel processing is enabled at runtime. Some BLAS-2, BLAS-3, LAPACK,

sparse solver, convolution, and FFT routines have been optimized for parallelism.

Controlling MLIB SMP parallelism at runtime Two methods are available at runtime to

specify the extent of parallel processing in MLIB.

1. Use MLIB_NUMBER_OF_THREADS, a shell environment variable that allows you to enable

parallelism within MLIB subprograms and to specify the maximum number of threads

that can be used in parallel regions.

Not setting MLIB_NUMBER_OF_THREADS has the same result as setting it to 1; parallel

processing is disabled within MLIB subroutines. Setting MLIB_NUMBER_OF_THREADS to the

number of CPUs in the system (or greater) allows parallelized MLIB subprograms to use

as many CPUs as are available to the application.

The following command lines show the C-like shell syntax and Bourne-like shell syntax to

use when setting the variable to two processors:

For C-like shell:

% setenv MLIB_NUMBER_OF_THREADS 2

For Bourne-like shell:

% export MLIB_NUMBER_OF_THREADS=2

MLIB_NUMBER_OF_THREADS is examined on the first call to a parallelized VECLIB

subprogram to establish the default parallel action within VECLIB.