An understatment similar to claims that Rubik's Cube have "more than 3 million combinations."
If the acceleration is constant, the average velocity is half the total velocity, or 17 km/s. At that speed, it would take about 290 nS to go 5 mm - a bit under a third of a millionth of a second.
Ten to the tenth g is a lot. I would guess that the maximum acceleration occurs quite near the end of the interval; more like an acceleration pulse. I don't know the mechanism, but a such a non-linear acceleration might stretch the time out a bit.