When the [data dependence analyzer||Data Dependence Analysis] is able to determine that a loop does not carry data dependences, the loop can be marked as parallel.
Based on the Bernstein conditions (RW, WW, WR produce a dependence in the loop, RR can be executed in any order) the data dependence analyzer determines a partial execution order of the iterations in a loop. When the loop does not contain RW, WW, WR tuples of memory accesses, the loop is called parallel, and its iterations can be executed in any order, in particular they can be executed in parallel.