The physical limits of current integrated circuits technology have been reached, and the performance of electronic devices can only be improved by replicating and adding processing components. This has led to a major concern in computing nowadays: the power consumption. To mitigate this fact, a major trend arised pointing towards making hardware (HW) more sustainable. Software (SW) plays a major role in computing sustainability as it is who exploits HW, ultimately determining its behavior. Despite that, the field of SW sustainability is in its early stages yet. Producing sustainable code is a cumbersome and error-prone task, requiring expert hands that often achieve partial success. This scenario asks for new robust technologies for enabling the design of sustainable SW to create a synergy with the new HW developments towards major energy savings.
GENIUS will develop original methods to produce sustainable and efficient SW in an automatic way, by modifying its back-end assembly code to adjust it to any underlying architecture. The new cutting-edge methodologies developed will be versatile, efficient, and practical, as opposed to the limited or very specific existing solutions.
The most remarkable feature of GENIUS is the automatic redesign of applications into novel efficient and sustainable ones through the use of AI, without any hardware modifications. Its successful accomplishment will imply major benefits for science, industry, and society. GENIUS allows, for instance, enlarging the battery life of an smartphone in a non-intrusive way, just by compiling its Apps with the proposed technology (without access to their source code). However, the significance of the project goes far beyond that, given that most of the existing trends for the future assume the massive presence of sensors and computing devices in society (e.g., Internet of Things, Smart Grids, or Smart Cities), boosting the ever raising presence of (mostly low-power) electronic components around us.

In order to be able to carry on these researchs a multidisciplinary team have been formed. Some of them are part of GOAL research group at Universidad de Cádiz.

Acknowledgements: This work was supported by the Junta de Andalucía and ERDF (GENIUS — P18-2399).