No question the E-M5 is faster. The X-A1 has slower AF than my X-Pro1 or the X100S, just a bit. The EM-5 is still has slightly faster AF than the X-Pro1 with the latest firmware update, but not by any practically meaningful difference.
I rented an X100S for a couple weeks on vacation, and actually, I think the AF on my X-Pro1 is faster than the X100S. Just my 2¢.
With all due respect, I have the E-M5 and also the X-M1, X-E1, and X100. The E-M5 has the fastest autofocus. And, quite frankly, if you figure that the X-E1 and X-Pro1 (with firmware upgrades) have similar AF speeds, the E-M5 still beats both of them. It is a practically meaningful difference (not insignificant). I have done comparative autofocus shooting with these cameras.
I just received the X-T1 which has faster autofocus, but still not faster than the E-M5 (or E-M1).
(BTW, I have also shot at Mare Island with my cameras.)
caver3d
You can't track anything at 6 fps, as the mentioned cameras do not feature tracking AF in continuous mode. Focus is fixed before the first frame of the series and remains the same throughout the entire series (so do exposure, DR and white balance). It's not a question of semantics or getting hung up, it's a mere question of facts: If there is no autofocus (because the camera is not focussing), there is no AF performance to discuss. You could just as well have used manual focus and prefocussed on the pit. Effectively, that's what AF-C did when it focussed the camera once(!) before taking the first shot of the series.
Whether the frames are all in focus (or not) is a matter of depth-of-field (as a function of aperture, focal length and circle of confusion) and the point the camera was focussing before the first shot of the series was taken. It's not a matter of AF speed, because there was no AF operation (and hence no AF speed) between shots of this series.
Of course, none of this is related to the X-A1, which features a simple, non-tacking CDAF. Not really fast, but fast enough for most. Of course, with the proper technique and experience, you can pretty much shoot anything with anything.
The jury is still out on the performance of the PDAF featured in the latest Olympus models. Reviews have been inconclusive, even DPR didn't actually test PDAF tracking in their EM-1 "in-depth review" expect for a few demo shots at an official Olympus event that are basically worthless. This is strange, as Fuji's predictive PDAF tracking is tested all the time, and I (and others) have the images to prove how well it works.