This project aims to…
- Define a methodology to combine static and probabilistic Worst Case Execution Times (WCET) analysis
- Static analysis (STA) is safe but complex – so expensive – to implement.
- Measurement-based timing analysis (MBTA) methods, and more specifically probabilistic MBTA are relatively easy to implement but a high confidence level is hard to obtain.
- The goal of C-PASTA is to investigate how to decrease the STA modeling effort and increase the confidence level on MBPTA estimations by combining the two methods.
- Towards that goal, we propose (i) to identify the architectural elements that contribute the most to the execution time, (ii) to focus the STA modeling effort on those elements, (iii) to leverage the HW model used by STA and the SW analysis performed by STA to improve the measurement coverage of MBPTA.