Abstract
Even though three-dimensional (3D) displays have been introduced in relatively recent times in the context of display technology, they have undergone a rapid evolution, to the point that a plethora of equipment able to reproduce dynamic three-dimensional scenes in real time is now becoming commonplace in the consumer market. This paper's main contributions are (1) a clear definition of a 3D display, based on the visual depth cues supported, and (2) a hierarchical taxonomy of classes and subclasses of 3D displays, based on a set of properties that allows an unambiguous and systematic classification scheme for three-dimensional displays. Five main types of 3D displays are thus defined -two of those new-, aiming to provide a taxonomy that is largely backwards-compatible, but that also clarifies prior inconsistencies in the literature. This well-defined outline should also enable exploration of the 3D display space and devising of new 3D display systems.

Abstract
Precomputed Radiance Transport (PRT) was introduced as a technique to enable interactive navigation and distant environmental real time relighting of rigid scenes. Evaluating radiance transport is, however, a computationally very demanding task, which precludes PRT's utilization during the model design phase, since the user must wait for long periods of time before being able to light and navigate within the model. This paper proposes and validates an approach to provide visual feedback to the user as soon as possible, within PRT context. By resorting to parallel processing and progressive refinement, the user is quickly presented with a lower lighting resolution of the virtual model. This is then progressively refined by incrementally increasing the number of incident directions taken into account on transport computations. PRT is, however, a complex algorithm that requires frequent collective communications of huge volumes of data, thus constraining the maximum achievable speedup on a parallel system. This issue is analysed and an alternative workload distribution is proposed and evaluated on a 12 node dual processor cluster. The final solution ensures a good resource utilization rate, reducing response times from dozens of seconds to a few hundred milliseconds. Copyright © 2006 by the Association for Computing Machinery, Inc.

Abstract
The ability to interactively render dynamic scenes with global illumination is one of the main challenges in computer graphics. The improvement in performance of interactive ray tracing brought about by significant advances in hardware and careful exploitation of coherence has rendered the potential of interactive global illumination a reality. However, the simulation of complex light transport phenomena, such as diffuse interreflections, is still quite costly to compute in real time. In this paper we present a caching scheme, termed Instant Caching, based on a combination of irradiance caching and instant radiosity. By reutilising calculations from neighbouring computations this results in a speedup over previous instant radiosity-based approaches. Additionally, temporal coherence is exploited by identifying which computations have been invalidated due to geometric transformations and updating only those paths. The exploitation of spatial and temporal coherence allows us to achieve superior frame rates for interactive global illumination within dynamic scenes, without any precomputation or quality loss when compared to previous methods; handling of lighting and material changes are also demonstrated.

Abstract
The efficient execution of irregular data parallel applications, on dynamically shared computing clusters, requires novel approaches to manage the runtime load distribution. Stich environments have an unpredictable dynamic behaviour both due to the application requirements and to the available system's resources. This uncertainty was the main motivation to propose and evaluate an application level scheduler, where decisions are efficiently taken with improved accurate predictions on the environment's current and near future state, based on available incomplete and aged measured data. Bayesian decision networks are used as the scheduler's decision making mechanism: its effectiveness to manage the load distribution of a parallel ray tracer is assessed and compared with alternative strategies. The evaluation results, with complex scenes on a 7 shared nodes cluster with dynamically variable workloads, show considerable performance improvements over blind strategies, and stress the benefits over a sensor based deterministic approach of identical complexity.

Abstract
The efficient execution of irregular parallel applications on shared distributed systems requires novel approaches to scheduling, since both the application requirements and the system resources exhibit an unpredictable behavior. This paper proposes Bayesian decision networks as the paradigm to handle the uncertainty a scheduler has about the environment's current and future states. Experiments performed with a parallel ray tracer show promising performance improvements over a deterministic approach of identical complexity. These improvements grow as the level of system sharing and the application's workload irregularity increase, suggesting that the effectiveness of decision network based schedulers grows with the complexity of the environment being managed.

Abstract
This article presents an evaluation study of point-to-point and collective communication performance on a parallel processing system, a 16 node Parsytec PowerXplorer, using three different communication environments: PARIX, PVM and MPI.