Calculating huge phylogenetic trees based on the maximum likelihood method is computationally expensive. We introduced and implemented Subtree Equality Vectors in (P)AxML to accelerate topology evaluation. (P)AxML scales well on PC architectures obtaining run time accelerations between 51% and 65% over (parallel) fastDNAml for large alignments. We introduce additional algorithmic optimizations and novel distance-based heuristics, which further accelerate (P)AxML by 14% to 19% and 4% to 8% respectively. Furthermore, we present an alternative tree-building algorithm and a new randomized approach for inferring huge trees, which yielded improved trees and accelerated program execution at the same time by a factor greater than 6.