Winning teams: performance engineering during development

In 1995 Raytheon Systems initiated a program to develop a DSP-based array processor that would be used to process large amount of raw data from a sensor subsystem. However, because the next generation target processor had not yet been released the software development team had to complete a significant amount of the system and software development prior to the availability of the actual hardware. To foster strong algorithmic performance and efficient implementation in this high-risk development environment, they incorporated software performance engineering into their design methods. Loosely defined software engineering is a set of techniques designed to gather data, construct a system performance model, manage the risk of uncertainty, evaluate alternatives and verify the models and results. The author explains his experience on this project detailing the areas of SPE that were used throughout the life cycle of this software development program: execution modeling, input modeling, gathering of performance data, tools, design principles and software engineering.