I have observed this also, and technically, I think the slow speed of the encoders in relative mode is dependent on the resolution of the VST knob it is controlling, and not so much an issue the design of the MiniLab mkII hardware/software.

For example #1, let's say we have a VST knob that ranges from 0 to 127 in steps of 1.

In absolute mode, encoder=0 equates to knob=0 and encoder=127 equates to knob=127, and each step of the encoder increments (or decrements) the knob by 1 step. Thus it takes 127 steps of the encoder to go from the minimum knob value (0) to the maximum knob value (127).

In relative mode with no acceleration, each step of the encoder increments (or decrements) the knob by 1 step, which results in the same encoder speed as in absolute mode.

With maximum acceleration, each step of the encoder increments (or decrements) the knob by 3 steps (I think), so it only takes roughly 43 steps of the encoder to go from the minimum knob value to the maximum knob value. That's actually faster than with absolute mode.

For example #2, let's say we have a VST knob that ranges from 0 to 127, but this time in steps of 0.1.

In absolute mode, encoder=0 equates to knob=0 and encoder=127 equates to knob=127, and each step of the encoder increments (or decrements) the knob by 10 steps (which is a numeric change of 1 to the knob value). Thus it still takes 127 steps of the encoder to go from the minimum knob value (0) to the maximum knob value (127), so the encoder speed is the same as with example #1. But now each step of the encoder is worth 10 steps of the knob, so you can only set the knob to integer values, even though the knob has resolution of 0.1.

In relative mode with no acceleration, each step of the encoder increments (or decrements) the knob by 1 step (which is a numeric change of 0.1 to the knob value). It now takes 1270 steps of the encoder to go from the minimum knob value (0) to the maximum knob value (127), which results in a 10X decrease in knob speed compared to absolute mode for this same knob. But the advantage is that you can set the knob to any value (not just the integers). You just have to turn the encoder a lot more to go from 0 to 127.

With maximum acceleration, the knob is faster, but it still takes roughly 424 steps of the encoder to go from the minimum knob value to the maximum knob value. That is still roughly 3X slower than using absolute mode. So when a VST knob is divided into a large number of steps between the minimum and maximum values, relative mode will be slower than absolute mode -- there is no way around this.

Technically, neither mode is ideal. Relative mode provides the ability to set the knob to any possible value, but comes at the expense of slower knob speed when the knob has a large number of steps between the minimum and maximum values. Absolute mode provides consistent knob speed, but comes at the expense of lower knob resolution (only 127 positions between the minimum and maximum values).

I think the solution would be to add even higher acceleration factors for relative mode, to move the knob 10, 20, or even 100 steps for each step of the encoder. That way you could quickly move a knob that has hundreds (or even thousands) of steps between the min/max values.