I have been investigating the dtime_% fields in the diff and chip stages for gpc1 for the month of October (and also September, see [wiki:ProcessingTimesSept2010]).  The tag changed on 2010-10-29, which included a number of changes to the diff stage.