It is presented the impact brought by the new multicore and manycore systems in the computer engineering field. There were explained the limits of the actual mono-processor paradigm that involved the multicore shift. A state of the art in multicore architecture and compiler research was presented. We identified and analysed some important research challenges in multicores’ research and development, like the following: homogeneous vs. heterogeneous multicores, synergistic exploitation of parallelism grains, cache hierarchy, cache coherence and consistency protocols, interconnection networks, parallel programming models, applications’ parallelisation methods, simulation, benchmarking, automatic design space exploration, speculative multicore architectures, power consumption and thermal effects. There are necessary important progresses related to all these scientific challenges, in order to adequately develop and program multicore and manycore systems. The research approach must be a holistic one for each of the above topics. It is shown how these systems would change the computer science and computer engineering paradigm.