In my short time using the Garmin I've had 2 firmware updates. The first when I set it up and the second last Saturday. Don't actually know if the update affects the calculations you mentioned.
The numbers you stated, (9.3 & 12), are these SD from a Garmin and LR? If so, who is to say which is actually correct? The comparison between the two isn't valid. The only way to accurately measure accuracy is to measure each against an industry standard chronograph. Only this measurement will provide a usable comparison. If you don't, all your getting is data output that has no basis in accuracy. That can only come from testing to a standard, not against a comparison of chrono to chrono out there. If folks continue to test one make against the other, the only result you get will be a disparity between the two chronos and BOTH may be wrong. Or maybe one will be right. It's a coin toss.