We survey parallel programming models and languages using six criteria
to assess their suitability for realistic portable parallel programmi
ng. We argue that an ideal model should be easy to program, should hav
e a software development methodology, should be architecture-independe
nt, should be easy to understand, should guarantee performance, and sh
ould provide accurate information about the cost of programs. These cr
iteria reflect our belief that developments in parallelism must be dri
ven by a parallel software industry based on portability and efficienc
y. We consider programming models in six categories, depending on the
level of abstraction they provide. Those that are very abstract concea
l even the presence of parallelism at the software level. Such models
make software easy to build and port, but efficient and predictable pe
rformance is usually hard to achieve. At the other end of the spectrum
, low-level models make all of the messy issues of parallel programmin
g explicit (how many threads, how to place them, how to express commun
ication, and how to schedule communication), so that software is hard
to build and not very portable, but is usually efficient. Most recent
models are near the center of this spectrum, exploring the best tradeo
ffs between expressiveness and performance. A few models have achieved
both abstractness and efficiency. Both kinds of models raise the poss
ibility of parallelism as part of the mainstream of computing.