Heuristics for Scenario Creation to Enable General

Abstract-Embedded system applications can have quite com- prohibits global loop transformations. To go beyond the scope plex control flow graphs (CFGs). Often their control flow of the SCoP, and enable far more global optimizations, we prohibits design time optimizations, like advanced global loop could consider paths of the CFG in separation. However, transformations. To solve this problem, and enable far more global optimizations, we could consider paths of the CFG in coding al paths sepa ate wo caue a tremous ceisolation. However coding all paths separately would cause a copying. In practice we have to make a trade-off between tremendous code copying. In practice we have to trade-off the the extra optimization opportunities and the code size. To extra optimization opportunities vs. the code size. make this trade-off, in this paper we apply the system scenario To make this trade-off, in this paper we use so-called system technique [12]. This technique bundles similar control paths, scenarios. These scenarios bundle similar control paths, while.t.a. n still allowing sufficient optimizations. The problem treated in whe stilal in scent optizatinstthis paper is: what are the right scenarios; i.e., which paths The systematic scenario concept was first used in [14] to should be grouped together. For complex CFGs the number capture the data dependent dynamic behavior inside a thread, of possible scenarios (ways of grouping CFG paths) is huge; in order to better schedule a multi-threaded application on a it grows exponentially with the number of CFG paths. Therefore heterogeneous multi-processor architecture. In [8] scenarios heuristics are needed to quickly discover reasonable groupings. The main contribution of this paper is that we propose anchave been used to refine the estmaton of the Worst Case Ex