Energy Consumption and Runtime Performance Optimizations Applied to Hyperspectral Imaging Cancer Detection
Eduardo Juárez, Raquel Lazcano, Daniel Madroñal, C. Sanz · 2022 23rd International Symposium on Quality Electronic Design (ISQED) · 2022
The development of systems where functional and non- functional requirements need to be balanced is still an open research challenge. In this context, the goal of this work is two-fold: first, proposing a Y-chart dataflow-based design methodology to include energy consumption within the non- functional requirements to be optimized at design time and runtime; secondly, this Y-chart dataflow-based design has been combined with the optimization potential offered by the polyhedral model to optimize applications throughput. For verification purposes, a hyperspectral imaging classification algorithm to locate human brain tumor boundaries during surgical procedures has been selected as test-bench.Regarding the energy-aware optimization, promising results have been obtained in both design time and runtime loop implementations, as they have been able to reduce the system energy consumption in up to a 25.7%. In turn, the runtime performance optimizations results show that the largest speedups are reached when combining the potential of dataflow and polyhedral tools, with speedups close to 4× when compared to a sequential version.