The goal of PGMT is software affordability through architecture independent programming for parallel processors.
PGMT is an effort to design and build software tools that allow PGM to be affordably and efficiently used on a wide range of parallel processors so that applications, once built, for a parallel processor could be reused with little cost. The design will make it easy to translate an engineer's design of a weapon system into software for the weapon system's computers. By design, this approach is specifically applicable to parallel distributed - computer processing.
For the application software developer, PGMT provides:
Earlier and less costly development of new parallel processor software;
Faster modernization of parallel-processor software;
Lower costs for modernizing software;
Lower software maintenance costs: and
More uniform capabilities across parallel-processor architectures with more compatible software for various hardware configurations.
For the Navy, PGMT:
Accelerates the beneficial infusion of advanced HPC technology into Navy Research and Development;
Lowers costs, and makes it easier to develop and upgrade weapon systems;
Lowers the number of noncombatant personnel needed to maintain weapon systems;
Develops more-flexible C3 systems in less time; and increase the capability for embedded training and simulation.
Graphical description of multi processor architecture (flow graph oriented) minimizes learning curve due to similarity with PGM;
Easy to use MacIntosh-like user interface for all encapsulated tools;
Migration path for existing applications;
Generation of optimized C++ and C source code for target multiprocessor:
Primitives and command programs written in ANSI C depending upon Navy project requirements;
PGMT written in C++, as state-of-the-art object oriented language for representing complex systems.
Source code generated and then compiled with vendor supplied compilers for target multiprocessor;
Static and dynamic assignment of application to multiprocessor resources;
Reduction, by a factor of 15, of the cost of developing PGM compliant environments for most future multi-processor configurations:Requires only MPI compliance, and C++ compilers;
Requires time-to-market of new applications which are currently processor limited or the porting of existing applications from obsolete to new equipment in a matter of months not years.