I like this setup as a test shot because of its extreme dynamic range.
Light colored siding in full sun to black objects in deep shadows.
The 645D is clearly showing its superior dynamic range here.
You certainly can see the pattern noise there in the 5Dmk2 shot when the shadows are pushed.
Part of the problem is the 5Dmk2 does not have as wide a dynamic range as the 645D.
I think that will become even more apparent with thorough testing.
So the signal to noise ratio is less and as you view the shadows with a reduced signal, the noise becomes visible.
Why will shooting 160 ISO remove the pattern noise?
As I understand it the pattern noise is do to mismatched amplifiers on adjacent pixel channels or green filters with differing optical transmission characteristics. Canon of course doesn't admit it exists. How will gathering less photons offset what is a physical problem inherent in the electro-optical system?
At ISO 160 you are merely shooting at ISO 200 and overexposing 1/3 stop.
In return for a tiny bit cleaner shadows you lose 1/3 stop headroom on the high end.
That's okay in the studio with a limited range of light, but not so okay outdoors with a huge range of light levels.
You could de-rate the 5D and shoot at ISO 50 and do better shadow wise, but that would also loose you 1 stop range on the high end. The highlights would start blocking up.