During the first decades of development of integrated circuits the performance of processors increased at an exponential rateas was predicted by Gordon E. Moore when he published his so-called Moore's Law in 1965. It was clear, however, that the increase in compute speeds of single processor machines would eventually be curtailed by physical constraints. The ever increasing demand to solve more complex and larger problems could thus in the long run only be met by harnessing the power of multiple processors by using these in parallel fashion.

As a need arose to stimulate the development of parallel computing technologies, the biennial International Conference series on Parallel Computing (ParCo) was started in 1983. Since then this conference series has played a stimulating role in furthering the development and use of parallel machines. The success of these conferences was continued by ParCo2015, which was held from 1-4 September 2015 in Edinburgh, Scotland, UK.

As was the case with all previous events, ParCo2015 attracted many notable contributions depicting present and future developments in the parallel computing field. The contributed papers illustrate the many different trends, both established and newly emerging, that are influencing parallel computing.

The number of processors incorporated in parallel systems have rapidly increased over the last decade, raising the question of how to efficiently and effectively utilise the combined processing capabilities on a massively parallel scale. The difficulties experienced with the design of algorithms that scale well over a large number of processing elements have become increasingly apparent. The combination of the complexities encountered with parallel algorithm design, the deficiencies of the available software development tools to produce and maintain the resulting complex software and the complexity of scheduling tasks over thousands and even millions of processing nodes, represent a major challenge to constructing and using more powerful systems consisting of ever more processors. These challenges may prove to be even more difficult to overcome than the requirement to build more energy efficient systems.

To reach the goal of exascale computing, the next stage in the development of high performance systems, fundamentally new approaches are needed in order to surmount the aforesaid constraints. Exascale computing holds enormous promise in terms of increasing scientific knowledge acquisition and thus contributing to the future wellbeing and prosperity of humankind. Such powerful systems are, for example, needed for executing complex simulations and large information processing tasks resulting from large-scale scientific experiments. It is therefore vital that the parallel computing community succeeds in overcoming the associated challenges.

Innovative approaches that can assist in solving the problems encountered with the development and use of future high performance and high throughput systems were suggested by a number of conference speakers. Thus, for example, the incorporation of learning capabilities into processors may form the basis for more intelligent systems that can more readily adapt to changing processing requirements. Improved automatic scheduling of processing tasks may lead to greater efficiency and make systems easier to program. More flexible hardware designs, such as are proposed by, for example, FPGAs offer further perspectives. Hardware could be made more reliable by improved monitoring, with automatic action taken if components fail.

This volume is a record of the stimulating exchange of information and innovative ideas to which all attendees contributed. In addition to the contributed papers, a number of mini-symposia were held, each focusing on one research theme. The topics covered by all contributors are reflected in the selection of accepted papers constituting these proceedings.

The organisers wish to thank all organisations and individuals who contributed to the success of this event. In particular the organisers wish to thank Pavlos Petoumenos for his assistance with producing this manuscript.

Gerhard Joubert

Hugh Leather

Mark Parsons

Frans Peters

Mark Sawyer

Date: 2015-12-15