A major challenge in protein-protein docking is the distinction between near-native and decoy complex predictions. It has been shown that near native solutions are usually located at the bottom of deep and densely populated funnel structures in the binding energy plot of the complex. Thus analysis, whether the energy plot of the vicinity of a docking solution is “funnel like”, can serve as a validation of a proposed solution. Generation of such densely sampled plots, however, is a major computational challenge. In the current work, we used GPUs to tackle this computational challenge. GPU (Graphic Processing Unit) is a dedicated hardware for parallel computations. Originally aimed to accelerate graphic operations, it consists of a grid of housands of cores which operate simultaneously, while each core executes the exact same command with different data. However its architecture is absolutely different from the serial CPUs. In order to exploit the full functionality of this highly parallel hardware the algorithms have to be thoroughly redesigned, which is a challenge by itself. We present FunnGen, an accurate and highly efficient and parallel algorithm for generation of the binding energy plots.