So I just figured something, it doesn't seem to exclusively be an issue of acceleration but one of velocity as well.
I bet if you lower your accel by half, it'll shake less.
It's really not so much the velocity, but how Mach3 accelerates and decelerates. Imagine driving in your car at a fairly quick rate, and then stopping very quickly, keeping even pressure on the brakes until you come to a complete stop. When you stop, the car will jerk backwards, as the suspension allows the inertia to carry it further than the tires.
Mach3 stops in a similar fashion, only instead of the suspension, it's the flex in the machine that allows it to jerk backwards as it comes to a complete stop. The bigger and heavier the machine, The more it's going to flex when you change direction. If you decelerate slower, then the machine may flex back to it's original state before it stops. If you decelerate faster, then it may not stabilize until it comes to a complete stop.
A lot of high end routers use proprietary controls, with S-Curve acceleration. At work, we have a Morbidelli router. It has a gantry that's supported at only one end, and is about 80" long. We regularly cut at 22,000 mm/min, and it cut's perfect, square corners at all times, and can take very deep cuts at high speeds with no noticeable flex. I don't know what the acceleration rate is, but I can assure you that it's far greater than yours. it reaches 22m/min in about 12" of travel.
However, if I try to cut something with lots of very short, straight segments, the quick direction changes can cause the gantry to vibrate enough to visibly affect the cut quality.
But in normal use, the S-Curve acceleration allows it to run perfectly smooth, with zero vibration.
On my homebuilt wooden machine, I've often used the Tempest test version of Mach3 for it's smoothness and accuracy, on a machine with a lot of flex.