I concur that it's unlikely that the program includes gravitational waves, or any other GR effects. So I'd guess that it only includes Newtonian gravity. Which I would think would be sufficient, as long as the distance of the BH away from the Earth >> the Schwazschild radius of the black holes, which is about 3km for a solar mass.
Why you are seeing a difference in heating is unclear. What I'd expect is that the tidal force would vary as 1/r^3 in the far field, and I think the displacement and heating would scale the same way, though I could be wrong about the scaling of the heating / displacement. I could also be incorrect in assuming that the distance was large enough to be in the far field.
So if there was a large difference in r, that could explain the difference. But you haven't given us any information on the distance or the ratio of distances. It's also possible that it has something to do with a rapidly oscillating displacement (you mentioned that the BH's were rotating around each other several hundered times a second), as others have mentioned. If the displacement calculations that estimated the power flows assumed slow changing fields, they might be way off, giving a large rapidly oscillating displacement that would overestimate the heating power.
If you can reach the author for comment it might give you some insight. I would not believe the answer from a game without working through the math in detail. Which would be a fairly arduous task.
That said, I was reasonably impressed when I looked up the integration algorithm,
http://universesandbox.com/forum/index.php?topic=13479.0, and found that the developer I said spent some time discussing the advantages of a symplectic integrator.
Possibly you could post to the same website above and a developer might have comments about how accurate their simulation was in the circumstances you describe.
One other point - if the simulation did include gravitational radiation, I'd expect the frequency of the BH's orbiting each other to vary with time, giving rise to a "chirp" signal as was seen for the inspiral LIGO detected. From your description, there was no such chirp signal, further suggesting that Gravitational waves were not modeled in the simulation.
I have not tried to estimate the details of the chirp - I believe there was some discussion of the "chip mass" in
https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.116.061102, the first Ligo detection paper. With enough effort you could try and estimate the time-to-inspiral for your BH scenario, and then get an idea of the power emitted in GW. I'm not sure where you'd get a good estimate of the Earth's efficiency at receiving GW's of such frequency. There have been papers that have looked at using the Earth as a GW detector that I've glanced at, I don't recall any of the details though.