Resilience of complex networks against attacks and efficient ways to dismantle a network with the less effort are two opposite aspects of network robustness. Both of them have many practical applications in social and technological networks. The problem of finding the best attack towards a network is of combinatorial complexity, thus methods of  intelligent attacks are necessary. In this contribution we perform all possible combinatorial attacks of a small number of nodes deletions (m=5) on networks of size N=100, and we analyze the resulting attacks according to the damage they produce measured by the size of the remaining largest component. We perform this in a series of networks of different and controlled modularities. Following, we compare the resulting statistics with state of the art targetted attacks, specially High Betweenness Adaptive (HBA), Module-Based attacks (MBA) and Collective Influence.