LowPowNoC - Evaluation and reduction of power dissipation in multicore systems based on Networks-on-Chip

Multicore chips are currently the norm in enterprise, scientific and desktop computing and have already made inroads into mobile and embedded computing. One of the major problems faced by multicore system designers is to achieve maximum performance and functionality while respecting the power dissipation budget, which limits how much energy the system can take from mains or batteries. The power budget should be set as low as possible, because of energy costs, impact on the environment, battery lifetime, cooling costs, among other factors. This project aims to investigate and develop design techniques to optimise multicores in such a way that they can perform the same functionality with the same level of performance, while dissipating less power. The main focus of the techniques addressed by this project is the on-chip communication infrastructure, which was introduced to allow the multiple processing cores to exchange data and can account for up to 30% of the total power dissipated by the chip. To maximise the impact of this research to the UK and European economy, the project will address system-level techniques which are available to local system design and integration companies, rather than technology-specific techniques that require control of the chip fabrication process, which is often outsourced.

The ultimate goal of this project is to reduce human's energy usage (and consequently carbon footprint). At least two percent of the total energy generated in the world today is consumed by computers. This means that every year, several hundreds of terawatt hour are consumed by desktop, enterprise, embedded and mobile computing, at a cost that approaches 100 billion pounds. Tackling that problem can be seen as a very bold goal for a small project that addresses a specific type of computing system, namely multicores based on networks-on-chip. There is evidence, however, that this particular type of computing system will be the dominant platform for enterprise, desktop and embedded computing within the next decade, so this is the reason for selecting this platform as the main focus of this project. The goal is that the advances in low power multicore computing achieved within this project could potentially result in savings at the order of millions of pounds, even for a rather modest adoption by industry. The ideal scenario, however, would be to benefit a large portion of the world's population by reducing spending on energy consumed by the computing infrastructure as well as on related infrastructure such as cooling and thermal insulation. Additionally, as the applications of computing to increase modern society's energy efficiency becomes more widespread (e.g. building automation, intelligent transportation systems, smart grids), the energy overhead of the computing infrastructure itself becomes critical to the success of those initiatives, therefore the need for solutions such as this one is even more evident.As academic researchers, the project team has no direct control over the energy demand of IT users, so the project focuses instead on the previous links of the chain that provide the users with IT infrastructure: IT systems industry, supported by the design automation and semiconductors industries. The immediate goal of the project is therefore to provide those industries with techniques and tools that can facilitate the design and production of power-efficient computing devices, which would then contribute to the achievement of the project's ultimate goal. The specific impact to those industries that can be directly affected by the project outcomes include cost reductions (for instance, by reducing battery capacity, heat insulation or cooling) and increased functionality (due to efficient use of the power budget) on the IT products they develop, which should increase their competitiveness in the global market. The project's focus on system-level techniques will potentially increase industry's design productivity and shorten its time-to-market, as it will enable evaluation and optimisation of power consumption at earlier stages of the design process.Finally, the project will have an impact to the academic research community, by proposing new low power techniques and by delivering an evaluation framework that can quantitatively evaluate the impact of those techniques to the power consumption in multicores based on networks-on-chip.