Next: With commuting directions
Up: Mapping Nodes to Processors
Previous: Tree-like
Here the data nodes are arranged in a linear list, with no distinct
sub sections that we can exploit by arranging them on different
processors. Instead, assuming p processors, we allocate the nth
element of the list to processor
. This does not
unfortunately exploit any locality in the list. Ideally, we would want
to know the length of the list and to divide it into p consecutive
chunks, and distribute one to each processor. We cannot do this as the
list is being created because there is no way of knowing the size of
the list in advance.
Timothy Lewis
11/12/1997