Congratulations are in order for the KNC/ORSOC team, and thanks to Bitcoinorama for sharing today's nailbiting rush for a hashing result before the "midnight deadline"! :-)
I'm quite confident Marcus will be able to tweak the firmware to where those HW errors are small and insignificant, and I'm pretty sure there is a bit more to come (I'm guessing the theoretical max speed is based on nominal voltage...) :-)
So, once again: Well done, to all involved!