I've had a Sigma 50 1.4 for about two years, which I bought primarily to use with my 5D classic as a large-aperture portrait lens. It was plenty sharp enough for me when the focus was accurate. Unfortunately, with the notoriously weak autofocus of the 5D classic, a fairly high percentage of my large-aperture portrait shots were out of focus.
I got a 5D Mark III last week and the first lens to undergo focus testing in my studio was that Sigma 50. I was relieved to see that it now focuses faster and more consistently. I think I dialed in +2 on the micro-adjust. Images were pretty sharp wide open and extremely sharp at f2.0.
I then tested my Canon 85mm 1.8 and my Canon 100mm 2.0. They did not need any micro-adjustment, but I did notice quite a bit of chromatic abberation wide open just in front of and behind the point of focus. Lightroom fixed it pretty easily and it went away when I stopped down two stops, but I did buy these lenses to shoot wide open, so I'll have to keep it in mind. I was shooting RAW at the time. I wonder if the Mark III would correct that if I selected JPG?
Anyway, whenever we talk about the autofocus performance of a particular lens, (fast, slow, accurate or innacurate) I think we should specify which camera body we are using it on.