Estimating Mobile Memory Requirements and Rendering Time for Remote Execution of the Graphics Pipeline

Abstract

Mobile devices have limited processing power, memory and battery power. Remote execution, wherein part or entire graphics pipeline is offloaded to a powerful surrogate server, is an attractive solution for low end mobile devices such as PDAs and cell phones that lack floating point units or GPUs. We have found that remote execution of floating-point-intensive pipeline stages such as transform and geometry operations can produce speedups of up to 10 times for a low-end mobile device. We introduce generalized pipeline-splitting, a paradigm whereby 15 sub-stages of the graphics pipeline are instrumented with networking code such that they can run either on a mobile client or a surrogate server. To validate our concepts, we create Remote Mesa (RMesa). As a foundation for deciding which stages of the pipeline would benefit from remote execution, this paper presents analytical models for the overall rendering time, memory requirements and roundtrip network delay incurred by RMesa.