IMHO, a lot of it has to do with how you're performing these tests. I've always found tests performed in bright light, during the middle of the day, at extremely high ISOs to be rather silly, since you'd never crank the ISO up that high during normal shooting situations. The most practical tests would be in low-light environments that actually require high ISO shooting.
I'm not curious enough about noise performance to setup some makeshift test, so I only push the ISO as the situations that requiring doing so present themselves during actual shoots. I only got the 5DIII up to ISO 3,200 thus far, and had no complaints with the results.
I have been using ISO 6400 on my 5D MK II in low light and getting motion blur from the slow shutter speeds, even at f/1.4, so I want to make sure I know what I can get away with before my next low light event.
However, to be fair, some will need fast shutter speeds when photographing indoor sports in what they call low light, and I think of as relatively good light, if thats how they will use the high ISO, its ok to check it out that way. Birders with 400mm f/5.6 lenses trying to capture images in the early evening will also need high ISO, even though we would not need it with our f/2.8 lenses.