Parallelism for Performance in AR Applications
Xiaolu Gu, C. Gough, Mark Billinghurst · UC Research Repository (University of Canterbury) · 2004
Augmented Reality (AR) applications are highly performance critical applications. Because they insert themselves between the user's senses and the real world, it is crucial that they run as fast as possible to provide maximum Realism. This requirement for high performance is offset by the fact that AR Applications are mathematically complex, making them inherently slow. Because AR applications generally form a pipeline of fairly distinct stages, they make good candidates for parallelisation. It should therefore be possible to maximise performance of AR applications by segregating these stages, forming a "waterfall" rather than a "pipeline" for the data to flow through. In this paper a multi-threaded AR application is implemented using the AR Toolkit, which allows each of the stages to attain their best possible performance and thus results in far greater overall performance and performance potential.