Understanding the High-Performance-Computing Community: A Software Engineer's Perspective

Abstract

Computational scientists developing software for HPC systems face unique software engineering issues. Attempts to transfer SE technologies to this domain must take these issues into account.

Software Development Cultures and Cooperation Problems: A Field Study of the Early Stages of Development of Software for a Scientific Community

“…The software is valued only to the extent that it supports the science, see also for example, Sanders and Kelly (2008), Basili et al, (2008).…”

Open science decoded

“…There is still a mismatch between some standard software-engineering environments and methodologies and the practice of working computational scientists developing parallel codes for high-performance computing systems 5 . However, as can be seen from the example of the ATLAS LHC experiment, significant parts of the physics research community certainly use many of the modern tools of software engineers to develop and manage their large code base.…”

Assessing climate model software quality: a defect density analysis of three models

“…In a study of high-performance computing (HPC) communities, Basili et al (2008) find that scientists value scientific output as the highest priority and make decisions on program attributes accordingly. For instance, an increase in machine performance is often seen as the opportunity to add scientific complexity to their programs, not as an opportunity to save on execution time (since that may not serve as great a scientific purpose).…”

“…A more in-depth study of defect histories will give us insights into the kinds of defects climate modellers have difficulty with, and how the defects are hidden and found. As well, we suggest detailed case studies of the climate modelling development done in a similar manner to Carver et al (2007), or Basili et al (2008).…”