Today's radio access technologies demand flexible, high performance and power-efficient computers. Software-Defined Radios offer a high degree of flexibility but their performance is limited in energy-constrained environments. Re¬configurable Computers exhibit a good trade-off between per¬formance and energy consumption. The plethora of recon¬figur¬able devices and architectures suggest using middleware with virtualization and resource management capabilities also in Re¬configurable Computers. Moreover, Cognitive Radios can take benefit of the middleware approach obtaining waveform resource demands, exploiting advanced resource management techniques. This paper compares two approaches for dynamic partial recon¬figuration and their suitableness for Cognitive Radio middle¬ware. Our reconfiguration approaches are novel in the sense that they use an intermediate memory as a buffer to isolate the bit¬stream downloading from the actual reconfiguration. Per¬form¬ance and resource occupancy is measured and analyzed for both approaches.