| dc.description.abstract | We consider a multiprocessor graphics architecture object-parallel if graphics primitivesare assigned to processors without regard to screen location, and if each processorcompletely renders the primitives it is assigned. Such an approach leads tothe following problem: the images rendered by all processors must be merged, orcomposited, before they can be displayed. At worst, the number of pixels that mustbe merged is a frame per processor. Perhaps there is a more parsimonious approachto pixel merging in object-parallel architectures than merging a full frame from eachprocessor.In this paper we analyze the number of pixels that must be merged in object-parallelarchitectures. Our analysis is from the perspective that the number of pixels to bemerged is a function of the depth complexity of the graphics scene to be rendered,and a function of the depth complexity of each processor's subset of the scene tobe rendered. We derive a model of depth complexity of graphics scenes rendered onobject-parallel architectures. The model is based strictly on the graphics primitivesize distribution, and on number of processors. \Ve validate the model with tracedata from a number of graphics applications, and with trace-driven simulations ofrendering on object-parallel architectures.The results of our analysis suggest some directions in design of object-parallel architectures,and suggest that our model can be used in future analysis of designtrade-offs in these architectures." | en_US |
