Parallel computing platforms have revolutionised the hardware landscape by providing high-performance, low-energy, and specialized (viz. heterogeneous) processing capabilities to a variety of application domains, including mobile, embedded, data-centre and high-performance computing. However, to leverage their potential, system designers must strike a difficult balance in the apportionment of resources to the application components, striving to avoid under- or over-provisions against worst-case utilisation profiles. The entanglement of hardware components in the emerging platforms and the complex behaviour of parallel applications raise conflicting resource requirements, more so in smart, (self-)adaptive and autonomous systems. This scenario presents the hard challenge of understanding and controlling, statically and dynamically, the trade-offs in the usage of system resources, (time, space, energy, and data), also from the perspective of the development and maintenance efforts.
Making resource-usage trade-offs at specification, design, implementation, and run time requires profound awareness of the local and global impact caused by parallel threads of applications on individual resources. Such awareness is crucial for academic researchers and industrial practitioners across all European and COST member countries, and, therefore, a strategic priority. Reaching this goal requires acting at two levels:
(1) networking otherwise fragmented research efforts towards more holistic views of the problem and the solution;
(2) leveraging appropriate educational and technology assets to improve the understanding and management of resources by the academia and industry of underperforming economies, in order to promote cooperation inside Europe and achieve economical and societal benefits.