The basis for discussions at the workshop was five central issues that were to be addressed if the end goal of realizing effective production quality HPC programming and execution environments is to be achieved. These are:
What are the characteristics of scientific and engineering applications and algorithms that require high-performance computer systems? What are the highest priority issues for system software and tools implied by these applications and algorithms? How will these applications and algorithms contribute to the long-term commercial viability of high-performance computing systems? What particular system software and tools will attract independent software vendors to participate in the commercialization of these applications and algorithms?
What types of systems software and tools are needed and can be provided to facilitate the creation, monitoring, and analysis of portable, scalable application software in high-performance computing environments? What would be required to motivate the user to use such system software and tools?
What are the constraints imposed upon system software designers by current and anticipated architectures? Are key developments in system software hampered because architectural support is lacking? What aspects of high-performance systems software are likely to place significant new requirements on future architecture?
What barriers prevent system software and tools developed by the research community from becoming products supported by system vendors and/or independent software vendors? What are the barriers to the use of these products by application developers? What should be the strategy for removing these barriers and explicitly what is the government's role in that strategy?