To compile a program for parallel execution using kapf
, use the following commands:
kapf /conc /cmp=myprog_mp.f myprog.f fort myprog_mp.f /fast/optimize=(tune=host,level=5) /automatic link myprog_cmp.obj/threads_enable/EXECUTABLE=myprog.exe,SYS$COMMON:[SYSLIB]kmp/lib
An explanation of the compiler and linker qualifiers follows:
/automatic - tells the compiler to allocate
local variables on the run-time stack.
/conc - tells KAP to restructure source code
for parallel processing.
/cmp - causes KAP to save the optimized
source program under the file name of your choice. The
kapf default is to name the optimized source
file_name_cmp .
/fast - sets the following qualifiers that
can improve run-time performance: /alignment=natural
, assume=noaccuracy_sensitive , math_
library=fast , and /optimize=level=4 .
/sys$common:[syslib]kmp.olb - tells the
linker to use the KAP parallel processing library for OpenVMS
Alpha. The parallel processing library provides an interface
to DECthreads. See Section 3.1 for more
information.
/threads_enable - tells the linker to
allow a multithreaded application to have a thread executing
on every CPU in a multiprocessor system. See Section 3.1 for more information.
/optimize=(tune=host,level=5) -
tells the compiler to optimize for the architecture of
the host processor. Note that the Digital Fortran
/optimize=(tune=host,level=5) qualifiers and the KAP
Fortran /tune qualifier work independently and
perform different optimizations. For information about the
KAP Fortran /tune qualifier, see Section 4.1.7.
kapf
to process a file, you must set the Digital Fortran compiler
and linker qualifiers appropriately. For this reason, Digital
recommends that you use kfort whenever possible,
as kfort automatically sets the compiler and linker
qualifiers correctly.