Heterogeneous computing systems increasingly leverage FP-GAs in the cloud and embedded use cases. With cloud FPGAs being remotely accessible, security is a critical concern. Recent studies show adversaries can exploit FPGA logic to create and remotely deploy malicious power-wasting circuits that consume excessive dynamic power, potentially injecting faults or causing denial of service. This work analyzes the most common reconfigurable power-wasting primitives to assess their power consumption, detection challenges, and attack effectiveness. We further propose new, logic-based, and resource-efficient variations of these circuits and experimentally evaluate them on two families of AMD FPGAs. Finally, we discuss factors influencing attack effectiveness and compare the studied designs' trade-offs.