acceleration of an inhomogeneous system

this is NOT a huge imbalance. you spend almost 60% of the time in “Pair” and the ratio between the min and the max time is ~1.4. there are multiple parameters, that you can base your balance adjustments on. and you can do it dynamically, as it is quite possible, that the distribution of particles changes significantly over the course of a run, so running the balance command once will be good for a bit, but not as good anymore after a while.

also, you already have increased your performance over the original speed by a factor of 4.6 and your original desktop of 8.2. that is quite a lot with one little change. how much more do you want?