I'll start off by saying that I'm not an expert by any means. Just a dude thats pretty passionate about both the arduino and DIY CNC!
For the sketch you see in the video:
I have the delay between pulses set to delay(200); this is because speed profiling is not setup and as everyone knows you can't just start a stepper at top speed
so in order to prevent stalling the motor out while running it with the NES controller, I opted to just run it at a medium speed. What you're seeing the Z and Y axis do is about 60-70% of their max rapid speed.
For the embedded program (grbl) that actually parses gcode:
At the current time, we have not experienced any stalling problems that were related to the Atmega168 not keeping up. Sort of amazing really.
Right now we have the duration of the step pulse set to 5 microseconds which figures out to a 100,000 pulses per second (with a 5 microsecond delay between pulses).