R: BESTmcmc is very slow with large n

I am running BESTmcmc from the r Package 'BEST' to compare two groups frpm A/B Testing.

I am wondering, how to determine the optimal n (=number of observations) as the algorithm is very slow.

The Dataset contains of 370k+ Observations. As I cannot wait days for results of this analysis I would like to find out what minimum number of observations I need to get reliable results.

I already tried with Power Analysis, which suggests n=40k. This however is still very very slow.

Any ideas?

Thanks in advance.