dataflower: harnessing heterogeneous parallelism for creative applications

Digital artists have always attempted to explore the state of the art capabilities of available computer hardware. Modern tools like Processing and OpenFrameworks, together with good online documentation and supportive communities, have made programming an approachable skill for every artist.

However, computer hardware is undergoing a paradigm shift. Pushed by ever increasing demands for performance and limited by ever increasing costs in power usage and data transfer speeds, hardware manufacturers are gradually shifting towards designs that combine many different and highly specialised processors. Taking full advantage of this power comes with the cost of abandoning the cognitive model of the single sequential processor. For artists, this means that their tools need to evolve.

dataflower is a novel collaborative visual development environment that makes high performance heterogeneous parallel hardware accessible for creative applications. We propose to exploit the potential of the visual dataflow programming paradigm to allow digital artists to intuitively design high performance applications, while abstracting the challenges posed by parallel heterogeneous computing systems.

This project's ultimate goal is fostering real-time experimentation, bridging the gap between beginners and experts, and improving communication and collaboration in multidisciplinary teams working in creative programming projects.

This project is being supervised by Professor Luis Paulo Santos (Minho University) and Professor Rui Rodrigues (FEUP)