ALMS wins, pretty much.
Though the EM-5 acquits itself fairly well, it still suffers the huge problem of all EVF cameras that it goes blank while shooting.
This makes shooting an action sequence in burst mode a lottery.
While I wouldn't normally advocate spray and pray shooting, it is a huge issue when trying to capture a car spinning out of control (for example).
The C-AF and C-AF[TR] modes are not as good as might be expected from the hype.
Expect about 1/2 second lock-on time for the continuous AF, so the first image in a burst is taken, then a gap, then the rest of the burst follows.
Burst speed is a lottery too, if using the focus priority (not release priority) to ensure the images are in focus.
Hah - allegedly in focus!
In SAF and CAF modes with Release priority off (so it should focus first and only shoot in focus) I have plenty of out of focus shots from today
Don't try to use it for BiF either, unless your panning is perfect and you never let the bird out of the focus rectangle. Oh and while acquiring the Bird in Flight the lens had better lock focus first time, because it can go so far out of focus that your subject simply vanishes in the EVF and you can't track it any more.
Strangely when trying to focus I got worse results tracking the subject with half pressed shutter button, and better simply mashing the shutter down while tracking in the centre focus point.
I have no idea how well it would work using an off centre focus point. Maybe I will get brave and try that tomorrow.
The EM-5 is better than the E-P1 and better than the GH-1 with the same lenses and same subjects. Burst mode is better, faster, and writes to memory quicker.
Oh and battery life of the EM-5 using the EVF measured in the thousands of shots (using burst mode and not chimping) rather than hundreds, I filled two 16G memory cards of RAW images on only one battery in a two hour session.
Overall I think it was worth it, compared to my existing micro 4/.3 cameras, as an upgrade for sports. But not in the same league as a DSLR, pretty much as I expected.