The Naive Approach

• Consider a simple program relating elements of two lists:

  vproc([],[]).
  vproc([H|T],[HR|TR]) :-
          process_element(H,HR),
          vproc(T,TR).

• Will assume called in the ``forwards'' way, with ground input.
• Parallelization using ``control parallelism'':

  vproc([],[]).
  vproc([H|T],[HR|TR]) :-
          process_element(H,HR) &
          vproc(T,TR).

• ``Fine grained'' operation:

  process_element(H,HR) :-
          HR is ((((H*2)/5)^2) + (((H*6)/2)^3)) / 2.

• Based on the fact that the iterations are independent (proved e.g. by the &-Prolog / CIAO compilers).
• Also determinacy information assumed.
• Similarly parallelized using ``data-parallelism'' (given the same information).