The LOCATION annotation specifies the processor on which the
annotated process is to execute. It is similar in form and function
to an array reference. It has the general form LOCATION(I
,
...,I
), where 
and the I
have the same form as
the indices in an array reference. The indices must not reference a
processor array element that is outside the bounds specified by the
PROCESSORS declaration provided in the process or subroutine in
which the annotation occurs.
The following code fragment shows how the program ring1.fm (§3.2.3) might be extended to specify process placement. The PROCESSORS declaration indicates that this program is to execute in a virtual computer with 4 processors, while the LOCATION annotation placed on the process call specifies that each ringnode process is to execute on a separate virtual processor.
==
program ring1_with_mapping
parameter (nodes=4)
processors(nodes)
...
processdo i = 1, nodes
processcall ringnode(i, pi(i), po(i)) location(i)
endprocessdo
end
=1.03
The program tree.fm shows the a more complex use of mapping
constructs. The process tree creates a set of 
(
a
power of 2) processes connected in a binary tree. The mapping
construct ensures that processes at the same depth in the tree execute
on different processors, if 
, where 
is the number of
processors.
==
process tree(locn, n, toparent)
intent (in) locn, n, toparent
inport (integer) li, ri
outport (integer) lo, ro, toparent
processors(16)
if(n .gt. 1) then
channel(in=li, out=lo)
channel(in=ri, out=ro)
processes
processcall tree(locn,n/2,lo)
processcall tree(locn+n/2,n/2,ro) location(locn+n/2)
processcall reduce(li,ri,toparent)
endprocesses
else
call leaf(toparent)
endif
end
=1.03