•Two
views for data parallel computation, .e.g. A := B+C@east as found in ZPL’s dense arrays
•
•Virtual Processors: Each processor is allocated the work of scalar data
values, that is, think of a situation where there are as many processors as array
elements
•
•Variable Points Per Processor: Each processor is allocated from 1
to n values, but usually many
–1
element is virtual processor case
–n
elements is the sequential case
–Algorithms
with this feature are scalable