Today’s life is heavily influenced by massive calculations which are done on ever faster supercomputer systems. Be it simulations to determine the perfect form for a new wind turbine, analyzation of big medical data, or your everyday Google Search or weather forecast. Each of these is done on massive supercomputing systems to cope with the shear complexity of the task at hand.

And with each day there are numerous new beneficial applications being developed to improve and enhance our daily lives, but are so complex that they can only be executed by supercomputer systems.



But scaling these supercomputing systems to the ever increasing demand on computation complexity is getting more and more difficult as the energy demand of these new systems is rising accordingly and with it the complexity of keeping the systems cool enough to work properly. Already 15% of our total electrical energy is required to power all the computers at home, at work and in data centers, and the number is growing rapidly.

It is therefore not possible to satisfy the rising demand for computing power while still keeping the costs for energy and especially cooling of these systems in a usable range. One solution for this dilemma is to include different kinds of specialized processor types to reduce the power requirements and therefore the cooling requirements. But embedding and using these heterogeneous types of processors in the best possible way is a pretty demanding challenge for an application developer and can only be done with an extensive knowledge of the hardware and multiple software languages.



To solve these issues, FiPS will setup a programming methodology, in which just a single programming language is used to write the supercomputing program. The final software is then analysed by the FiPS methodology to determine the best processor types and the program is automatically prepared for these processors. Additionally, the user receives a prediction about how fast the computation and communication will be for the program, and how much energy will be consumed. Then the implementation can be updated, trying to increase performance and/or energy efficiency even further.

FiPS will have an ecological impact by reducing energy demand (and thus carbon dioxide emission), but also an economic impact by cutting the major energy and cooling costs. Supercomputing will therefore become cheaper and thus affordable for many other applications for which it is currently unfeasible to use supercomputer systems.