This is a good question. The versions that Diego Darriba, Tomas Flouri, and Alexis Stamatakis are working on include their Phylogenetic Likelihood Library (PLL) described in this paper: 
http://sco.h-its.org/exelixis/pubs/Exelixis-RRDR-2013-2.pdf. I forked a version of their development code that just includes the optimized likelihood functions for SSE, AVX, and OpenMP. And that's what's in 
https://github.com/trayc7/FDPPDIV. So the version that includes the PLL is definitely faster. And I believe they have also implemented amino acid models. However, my development version has some new models for fossil calibration. At the moment, I recommend working with their versions since the fossil-calibration models have only been tested on a subset of realistic datasets. Ultimately, the new models I'm working on will be merged with their code.