Now at Syracuse University, Fox has set up a new program ACTION (Advanced Computing Technology is an Innovative Opportunity Now). This is funded by New York State to accelerate the introduction of parallel computing into the State's industry. The methodology is based directly on that proven successful in CP. The applications scientists are now in different industries-not in different Caltech or JPL departments. There are many differences in detail between the projects. The basic hardware is now available commercially and need not be developed concurrently with applications and systems software. However, the applications are much harder. In CP, a typical code was at most a few thousand lines long and often developed from scratch by each new graduate student. In ACTION, the codes are typically larger (say 100,000 lines) and longer lived.
We also find differences when we analyze the problem class. There are fewer regular synchronous problems in industry than in academia and many more of the metaproblem class with several different interrelated functions.
Table 19.1 presents some initial results of a survey of industrial applications [Fox:92e]. Note that we are at the stage analogous to the beginnings of CP when we first wandered around Caltech talking to computational scientists.
Table 19.1: An Initial Survey of Industry and Government Opportunities for High-Performance (Parallel) Computing
In general, we find that the central parallel algorithms needed in industry have usually already been studied by the research community. Thus, again we find that, ``in principle,'' parallel computing works. However, we have an even harder software problem and it is not clear that the software issues key to the research applications are the same for industry. As described in Chapter 14 for High Performance Fortran, software standards are critical so companies can be assured that their parallel software investment will be protected as hardware evolves.
One interesting initial conclusion about the industrial opportunities for parallel computers concerns the type of applications. Simulations of various sorts dominated the previous chapters of this book and most academic computing. However, we find that the industrial applications show that simulation, while very promising, is not the largest market in the long run. Rather, we live in the ``information area'' and it is in the processing of information that parallel computing will have its largest opportunity. This is not (just) transaction processing for the galaxywide network of automatic teller machines; rather, it is the storage and access of information followed by major processing (``number-crunching''). Examples include the interpretation of data from NASA's ``mission to planet Earth'' where the processing is large-scale image analysis; the scanning and correlation of technical and electronic information from the world's media to give early warning for economic and social crises; the integration of medicaid databases to lower the burden on doctors and patients and identify inefficiencies. Interestingly, such information processing is currently not stressed in the national high-performance computing initiative.