The results look really confusing to me in two ways. The first one is that I've seen recommendations to set effective_io_concurrency=256 (or more) on EBS.
I would not expect this to make much of a difference on a table which is perfectly correlated with the index. You would have to create an accounts table which is randomly ordered to have a meaningful benchmark of the eic parameter.
I don't know why the default for eic is 1. It seems like that just turns on the eic mechanism, without any hope of benefiting from it.
