extended scope of PATMOS is important
Why the power efficiency R&D landcape must be reconstructed
Until 2030 the electricity consumption of computers, the internet and other ICT infrastructures has been predicted to increase by a factor of 30, if current trends continue [1, 2, 3, 4]. During the past 11 years the price of crude oil has increased by a factor of 9. Growing cost of exploration and a coming shortage of resources will probably again increase oil and gas prices. If until 2030 by a factor of 10, this would mean a total electricity cost increase by a factor of 300.
This could have serious consequences on the overall afford-ability of computing and the internet. This is a dramatically critical survival issue of the world-wide computing and ICT infrastructures because power efficiency research results are dramatically insufficient. This problem cannot be solved by evolutionary development. We urgently need a disruptive reconstruction of the entire R&D landscape such, that power efficiency research will have high priority far beyond parallel programming as well as microchip design and architecture.
Meanwhile design space exploration is limited by power efficiency. Lithography of microelectronics is no more the key. Also because of leakage current Moore's Law is no longer the trump card. See here. Power-efficient computing will be based on parallelism and datapath structures.
We must break current trends. We have to bridge the chasms between abstraction levels, paradigm domains and programming models. We must introduce connected thinking for close cooperation with all application domains. We urgently must overcome their reductionist attitudes telling us that this is not their job. We cannot afford their Aristotelian point of view onto their special area as the center of the world. We must introduce connected thinking by a Copernican organization of the entire ICT-related R&D landscape.