In this paper, a real-time energy-optimal strategy exploiting preview information resulting from connectivity and autonomous vehicle (CAV) technology is verified by experimental analyses linked with on-board computing methods on a real vehicle-in-the-loop testbed with virtual road and traffic light system. An energy-optimal deceleration planning/following system (EDPS) as a service-oriented technology for electrified vehicles utilizing preview information is applied to a micro-controller, where the data access route and location on the given system architecture are optimized to shorten computing time. Also, two types of multicore strategies are comparatively analyzed to efficiently operate computing resources of the embedded controller as well as to distribute computing loads, and the strategy comparisons indicate that a function-level task partition considering target cores in advance can practically reduce computing loads. In the vehicle-in-the-loop simulations (VILS) with a realistic driving on the virtual road and CAV technology-based information, the embedded EDPS planning results illustrate that the energy-optimal speed profiles are properly computed on a commercial automotive microcontroller while stably processing real-time data input/output and optimal planning within the given time constraints.