Global Arrays' primary strength is its ability to manipulate two-dimensional matrices. Hence, applications that rely on those and on linear algebraic operations may be easier and more efficient to express in Global Arrays than in the other HPVM APIs.
In addition, Global Arrays has been ported to distributed-memory MPPs, scalable shared-memory systems, and loosely-coupled workstation clusters. So applications needing some degree of portability may be worth implementing in Global Arrays, as opposed to SHMEM or FM.
Go to the first, previous, next, last section, table of contents.