All are global shutter. Don't know the price of the sensors, but no one seems to be making cameras with these sensors. Apertus might be making one with the 12MP, called the Axiom. But who knows when that is coming out, it is just an idea right now.
ON Semiconductor has a line called Vita, it includes at the top-end, the Vita25k. A 25MP, 35mm (FF), 1:1 aspect ratio, 53fps, 10bit, global shutter. After some digging and questions, I found out it costs like $4300 just for 1 colour sensor, lol.
There is a sensor design floating around called the AM41. One company that makes it is Alexima. It is a 4MP, 4/3", 500fps, 10bit, global shutter. It can go up to 800fps if you drop down to 7bit.
Sometimes makes me want to give Computer Engineering a try just so I could make some cameras with these sensors!
^ I didn't know about that 20MP CMOSIS sensor, where can I read something about it? Because the 12MP one (which costs around $1K) looks great on paper, but I'm not sure the HDR mode is suitable for video, and without it all you get are 10 stops of DR. Maybe this one will be better in this regard? Also, I'm guessing all those other 10-bit sensors won't have amazing DR...
Yeah, I think the final frontier is global shutter and at least 13 stops of dynamic range. For now all the global-shutter sensors I've seen hit 11 stops --- which is totally usable, don't get me wrong. But it my limited armchair research, it seems that after about 13 stops, you're at the "I don't really have to worry about dynamic range as much." (Witness the Sony F3, is that just S-log that makes it looks so nice?) You still have to think about dynamic range, but not dance around it as much. Of course, ever more is welcome.
If it is 11 stops in Global shutter, than switching to Rolling shutter will give you even more, at least an extra 2 or 3. That is the benefit of having a Global shutter sensor.
In the CMV12000, the HDR mode works by setting different shutter speeds to the odd and even columns of the sensor. To increase DR by 5 stops, if you're shooting 1/50s, that means your "highlight recovery" columns will be shooting at 1/1600s. It is up to the camera developer to merge all this info, but you're bound to have motion artifacts in the highlights. Plus you really shouldn't think of all that space as normal DR, it's just for highlight recovery, if you expose the faces of your actors in that area, you may get the saving-private-ryan fast shutter effect. This is all theory, though: I'm not 100% sure about the downsides, since I've never played with one of these sensors. That may change in the coming months... Any engineers around here interested in developing a motion camera?
edit: it also has a piecewise linear HDR mode, but I'm still not sure how effective this HDR mode is either