This is a presentation of initial ideas on techniques that can be used in order to achieve a predictable execution time in presence of cache memory.