Parallel computing has become extremely important in today's IT world. Almost all computers are now parallel systems. To efficiently use them one needs to divide a program into tasks and to schedule them onto the processors or cores of the system. Theoretically this is addressed by task scheduling where a program is described by a directed acyclic graph, a so called task graph. The nodes represent the tasks and the edges represent the communications between the tasks. Algorithms are designed to find the best scheduling of this graph onto a given parallel system. Many algorithms have been proposed so far.
Recently, machine learning has been successful with other hard problems. This project aims to investigate how machine learning can be applied to task scheduling. An enabling factor for this is that recent research of the Parallel and Reconfigurable Computing (PARC) lab produced a large set of optimal schedules which can be used for training a machine learning network. The algorithms will be implement and evaluated in Java, using existing frameworks.
Undergraduate
Machine learning algorithm for parallel computing
None
Lab allocations have not been finalised