Today is the moment of truth and I am once again checking myself. How good was my so-called Frankenstein mod really? I received a second XFX card for retesting, but was fair enough to put its shroud on the already tested board including cooler without disassembling the card as such again. Thus, the measurement results for the cooling and the volume are then also 1:1 comparable, because we can elegantly exclude the chip lottery with it. So these are ideal conditions.