First, to extract a maximal (or maximum) level of system performance, the rendering engines are kept equally fully busy until the end of a rendering cycle. In the context of parallel computing, this issue may be referred to as load balancing, where multiple rendering engines are working to address problems of unequal complexity. One or more aspects are directed towards avoiding a situation in which one or more rendering engines completes its task(s) before other rendering engines, thereby sitting idly while the other rendering engines that may be operating on more difficult (or complex) parts of the problem continue to operate, resulting in diminished parallel speed-up.

In addition, it is appreciated that different computers hosting the rendering engines may not necessarily be homogenous. For example, the different computers hosting the rendering engines 130 of FIG. 1 may have different computing capabilities. Here, the different computers may employ a combination of processor types with different processor core-counts and/or different clock speeds. By distributing tasks according to one or more embodiments of the present disclosure, different computers having such different capabilities may be better accommodated.

In addition, it is appreciated that each rendering engine may begin rendering computations at a different time. According to at least one embodiment, the controlling of multiple rendering engines to perform computations (e.g., for a particular pixel) need not include controlling the rendering engines to begin performance of the computations at a same time (e.g., a same particular time). As such, according to one or more embodiments, synchronizing the rendering engines to begin rendering computations at a particular time is not necessarily required.