As it says in the title, the php parser tests run is 2.5x slower compared to the js run. For one, there are some differences. All env and config objects are recreated for every test on the PHP side (which simplifies the logic and the messiness the sharing causes in the core code) whereas on the JS side, there is some amount of reuse. But, I don't think that in and of itself explains the performance difference.
On my laptop, the 5min test run for parser tests is a productivity dampener and will probably lead me to push the full test run to jenkins. So, at some point, we should investigate this and see if we can speed it up.