Search

Modern mobile processors are constrained by their limited energy resource and demanding applications that require fast execution. Single core designs use voltage/frequency throttling techniques that allow the system to switch between performant and efficient configurations to address this problem. Heterogeneous multicores also need to decide on which core to run rather than just adjust voltage and frequency. Consequently, they remain an open subject in terms of near optimal design and operation. This thesis investigates the performance and energy trade-off when migrating between heterogeneous cores and presents designs that enable low overhead transitions between cores through a series of contributions. The first contribution is based on a novel methodology, that can directly compare the execution of heterogeneous cores. With it, an in-depth investigation of the effects on the memory system when migrating between such cores is conducted. The analysis reveals that heterogeneous multiprocessor system slowdown is asymmetrical. In-Order core performance relies on the memory subsystem, while Out-of-Order execution depends on accurate speculation. A proposed design minimises migration overheads in In-Order cores without making the design prohibitively complex to implement. This is achieved by only sharing the larger caches and the translation state. The second contribution is a branch predictor design that transfers state when a migration occurs between heterogeneous cores. The design eliminates the warm up for Out-of-Order cores by transferring only minimal state. This improves post migration accuracy, potentially enabling better performance and energy efficiency. Finally, security has become a major concern for shared or transferable components in multicore systems, namely the branch predictor. The third contribution in this thesis investigates mitigation techniques of recently discovered side channel attacks. The proposed design flushes all but the most useful branch predictor state, ensuring isolation with minimal performance loss.

Street lighting is a ubiquitous utility. It does not only illuminate the streets during the night but also helps to prevent crime and traffic collisions. However, to sustain its operation is a heavy burden both financially and environmentally. Because of this, several initiatives have been proposed to reduce its energy consumption. However, most initiatives are mainly aimed at energy conservation and have given little consideration about the usefulness of street lighting. A Streetlight Usefulness Model, an evaluation metric used to measure the usefulness of street lighting to road users, is proposed. Using StreetlightSim, a real-time co-simulation environment developed as part of this research, the energy efficiency and usefulness of six existing street lighting schemes have been evaluated. Their performances were used as baseline results which later justified the proposal of Traffic-aware Lighting Scheme Management Network (TALiSMaN). Simulation results show that TALiSMaN can achieve comparable or improved usefulness (> 90%) to existing schemes, while consuming as little as 1 – 55% of the energy required by existing schemes. To consider the limitation of ‘off-grid’ streetlights – those powered locally by renewable energy, TALiSMaN has been enhanced with an energy demand predictor to ensure that a limited energy budget can be used fairly throughout the whole night. This enhanced scheme is known as TALiSMaN-Green. Combined with knowledge of the amount of energy stored, and predicting sunrise times, TALiSMaN-Green modulates the lighting levels requested by TALiSMaN if the energy stored is predicted to be insufficient for an entire night. The results show that this scheme extends the operational lifetime of solar-powered streetlights from 2 to 16 hours. Evaluated with real traffic flow and solar readings, TALiSMaN-Green can maintain streetlight usefulness at 60 – 80% (mean = 73% with standard deviation of 9%). In comparison, the streetlight usefulness of TALiSMaN was reduced to below 30%.