A parallel application interface

This paper dcscribes the use of APL2 to implement parallel algorithms. API,2 includes high level constructions for support of coarse grain parallel computing. It is possible to move Paris of applications between machines of different architecture without compilation or linking. This allows the partitioning of an algorithm among available processors at run time instead of design time. The API,2 coopcrative processing interface allows the programmer to make use of any processor on a TCP/IP connected network. The possibilities for imbedded fine grain parallel support are discussed.