Effective static analyses have been proposed which infer bounds on the numberof resolutions or reductions. These have the advantage of being independentfrom the platform on which the programs are executed and have been shown to beuseful in a number of applications, such as granularity control in parallelexecution. On the other hand, in distributed computation scenarios whereplatforms with different capabilities come into play, it is necessary toexpress costs in metrics that include the characteristics of the platform. Inparticular, it is specially interesting to be able to infer upper and lowerbounds on actual execution times. With this objective in mind, we propose anapproach which combines compile-time analysis for cost bounds with a one-timeprofiling of the platform in order to determine the values of certainparameters for a given platform. These parameters calibrate a cost model which,from then on, is able to compute statically time bound functions for proceduresand to predict with a significant degree of accuracy the execution times ofsuch procedures in the given platform. The approach has been implemented andintegrated in the CiaoPP system.
展开▼
