2017 1st Workshop on AutotuniNg and aDaptivity AppRoaches for Energy efficient HPC Systems

High Performance Computing (HPC) has been traditionally the domain of grand scientific challenges and a few industrial domains such as oil & gas or finance, where investments are large enough to support massive computing infrastructures. Nowadays, HPC is recognized as a powerful technology to increase the competitiveness of nations and their industrial sectors, including small scale but high-tech businesses – to compete, you must compute has become an ubiquitous slogan. However, given the performance, power and energy envelops and the increasingly complexity of each computing node, the conventional compiler and code optimization process is no more adequate and there is the need to move to runtime many decisions and explorations of the vast design space. In this context, runtime adaptivity and runtime autotuning is of paramount importance. Furthermore, reaching exascale poses the additional challenge of significantly limiting the energy envelope, while providing massive increases in computational capabilities which may increase the importance of runtime optimizations.

This workshop intends to be an informal, privileged, forum to present and discuss new ideas, challenges, open issues, and trends regarding autotuning and runtime adaptivity in the context of energy-efficient HPC systems. This includes but not limited to the entire vertical stack ranging from the software components (programming best-practices, tools, compilers, runtime environments, operating-systems) to the hardware and firmware components at support of high level autotuning and runtime adaptivity mechanisms. The workshop intends to bring together practitioners, people from supercomputing centers, and researchers interested on those topics. It is foreseen that the workshop may contribute to encourage collaborations among the participants.