The high performance embedded computing software initiative: C++ and parallelism extensions to the vector, signal, and image processing library standard

The high performance embedded computing software initiative (HPEC-SI) program is developing a unified computation and communication application programming interface (API) and framework for high performance signal processing tasks on parallel computers. The goal of the program is to address the high cost of software in Department of Defense (DoD) systems by improving the portability and productivity of signal processing application development threefold, while improving performance by one half compared to current practices. This paper describes the motivation for the HPEC-SI program, its goals and approaches, and progress of the HPEC-SI Working Groups in extending the vector, signal, and image processing library (VSIPL) standard to C++ and transparent operation in parallel computing systems. The current draft version of the C++ extensions to VSIPL is described, and highlights of its advantages are considered. This paper also examines results from the Demonstration Working Group, and describes requirements and plans developed by the Applied Research Working Group for data parallel extensions to VSIPL and describes Development Working Group progress so far in developing parallel VSIPL.