Novel methods and tools are needed for the performance evaluation of future embedded systems due to the increasing system complexity. Systems accommodate a large number of on-terminal and or downloadable applications offering the users with numerous services related to telecommunication, audio and video, digital television, internet and navigation. More flexibility, scalability and modularity is expected from execution platforms to support applications. Digital processing architectures will evolve from the current system-on-chips to massively parallel computers consisting of heterogeneous subsystems connected by a network-on-chip. As a consequence, the overall complexity of system evaluation will increase by orders of magnitude. The ABSOLUT performance simulation approach presented in this thesis combats evaluation complexity by abstracting the functionality of the applications with workload models consisting of instruction-like primitives. Workload models can be created from application specifications, measurement results, execution traces, or the source code. Complexity of execution platform models is also reduced since the data paths of processing elements need not be modelled in detail and data transfers and storage are simulated only from the performance point of view. The modelling approach enables early evaluation since mature hardware or software is not required for the modelling or simulation of complete systems. ABSOLUT is applied to a number of case studies including mobile phone usage, MP3 playback, MPEG4 encoding and decoding, 3D gaming, virtual network computing, and parallel software-defined radio applications. The platforms used in the studies represent both embedded systems and personal computers, and at the same time both currently existing platforms and future designs. The results obtained from simulations are compared to measurements from real platforms, which reveals an average difference of 12% in the results. This exceeds the accuracy requirements expected from virtual system-based simulation approaches intended for early evaluation.
|Award date||14 Dec 2012|
|Place of Publication||Oulu|
|Publication status||Published - 2012|
|MoE publication type||G5 Doctoral dissertation (article)|