Things make sense now. I remember a while back Facebook invested in an experiment with PHP on PyPy. They didn't pursue it, even though it produced impressive results. It seems their own in-house JIT has better performance?
PyPy is taking a radically different approach from what HHVM is doing (and really from what almost all other dynamic language systems are doing), and it's a fascinating system. Part of what's exciting about it is that it seems like it should be applicable to other languages with less effort than most other JITs, and we wanted to understand its potential for a language like PHP. We asked Maciej Fijalkowski (hi, Maciej, if you're reading!) to help us do a research prototype to see what the first few roadblocks would look like, and Maciej did a great job. Just because we didn't scrap our current project and shift all our resources to PyPy should not be seen as a negative reflection on PyPy at all.