This paper describes the application of statistical analysis of the timing behavior for a generic real-time task model. Using specific processor of ARM Cortex-A series and an empirical approach of time values retrieval, the algorithm to predict the upper bounds for the task of the time acquisition operation has been formulated. For the experimental verification of the algorithm, we have used the robust Measurement-Based Probabilistic Timing Analysis based on the Extreme value theory (EVT). In the ongoing work, we provide a systematic method for safe worst-case execution time estimations of programs running on the specific hardware. We focus here primary on the consequences of EVT assumptions and their correct interpretation for the upper bounds prediction.