panpvmc
- compile script for PVM programs on top of Panda
SYNOPSIS
panpvmc
[ options ]
source files
DESCRIPTION
This command compiles a PanPVM program. PanPVM is a (nearly complete)
implementation of PVM3 on top of Panda; the features of PVM that
are not implemented are dynamic process management and persistent
groups. PanPVM programs run in SPMD style. See
panpvm(3).
It is built on top of
panc(1),
from which it inherits all architecture options.
panpvmc adds a number of PanPVM-specific options.
OPTIONS
panpvmc control options:
- -n
-
display commands to be executed, do not run them
- -verbose
-
increase panpvmc verbosity.
Language options:
- -f
-
Invoke
g77
as the compiler/linker.
For source files whose name ends in .f or .F,
this is the default. For all other source files,
gcc
as invoked as
compiler/linker.
Compilation options:
- -O, -O2, -O3
-
Produce optimized code. Links optimized libraries, from which statistics
options have been compiled out. By default, a non-optimized library with
statistics compiled in is linked.
- -pg
-
Produce profiling code, see
gprof(1).
Warning: each process creates a file
gmon.out
in the current directory. These files interfere generally. Under
Solaris, this can be solved be setting environment variable PROFDIR, under
BSDI/OS on DAS an option -pg is provided by prun, which causes all processes
to chdir to a separate directory. This will cause confusion if a parallel
process wants to open a file in the current directory, since the latter
has changed.
SEE ALSO
panc(1),
prun(1),
panpvm(3)
Markup created by unroff 1.0, December 01, 1998.