Hi, interesting results.
However, I have some difficulties in explaining this from a physical point of view.
Let's assume power output is proportional to mass flow trough the engine.
And, that mass flow is (among displacement and rpm) a function of 1:volumetric efficiency, 2:manifold pressure, and 3:manifold temperature.
For the sake of argument, let us assume 1 and 2 remain unchanged (in practice boost will differ if intercooler restriction differs and the WG is plumbed directly from the compressor housing).
A decrease in power of 13% (ebay vs garrett) roughly corresponds to a inlet temperature increase of 40 degrees celsius (ideal gas law).
I have been doing quite some data logging with the ebay IC, and although I have not been running huge amounts of boost, I have not seen higher than 20 degrees above ambient at max.
Also, when swapping the 940 cooler for the ebay cooler (12x24) my boost increased by 0.2-0.3 bar. Yet when direct swapping a stock one for a ebay cooler, you lose power?
How did you control boost? Manifold pressure the same for all runs?
Edit: Ok, some more careful reading, so you see less boost with the ebay core vs the garret. Is this also the case for the stock 940 one? Because my results have shown the opposite. Maybe it's a different cooler, although it looks the same.