A system-level exploration method called architectural probing is proposed. It embodies the widely accepted concepts of behaviour/architecture codesign, and allows one to selectively gather functional information from a functional diagram and feed it into an architectural diagram, thus enabling a reduction in the amount of details that are simulated at the architectural level. The effectiveness of this method is demonstrated in a case study that involves the exploration of data queuing management schemes for a packet switch system. As shown in the experimental results, the method enables the evaluation of the impact of different data management algorithms and different system parameters such as DMA size and packet payload size, in terms of power and performance tradeoff by estimating the average energy-per-bit-transferred, on each architectural component, such as memories, buses and bus arbiters, involved in the implementation of the queuing management system.