The automotive industry is facing many challenges in its transformation towards the Software Defined Vehicle (SDV). The number of software components and their complexity is rising and the real-time issues increase with it. In this context, there is a need to accelerate the integration of mix-critical components with hard and soft real-time properties on complex hardware platforms. We propose in this paper a performance prediction approach that allows to determine the feasibility of real-time constraints early in the design cycle to detect and fix timing issues as early as possible. The novelty of our approach is that it automates much of the constraint definition work and reduces pessimism in the feasibility evaluation.