Sadly, I've mostly come to the same conclusion in that I at least need a MIP Gap to tell me how far I am from the optimal solution.

My only issue with LP and MILP solvers is scalibility. You have none! Multicore doesn't really seem to help any.