Dynamic Application Autotuning for Approximate Computing
Content of the lecture:
Several classes of applications can expose at runtime a set of software knobs (including application parameters and code transformations) to trade-off the quality of the results and the throughput. Resource management and application autotuning are key issues for enabling computing systems to adjust their behavior in the face of changing conditions, operating environments, usage contexts and resource availability while meeting energy-efficiency and Quality-of-Service requirements.
This lecture will present dynamic autotuning techniques for the multi-objective optimization of applications to tune software knobs in an adaptive scenario to trade-off accuracy versus performance. Machine learning techniques can be used to predict the system behavior based on a set of training data. The main challenge is to exploit design-time and run-time concepts to lead to an effective way of “self-aware” computing.