Thanks but dont think this is it. With the blockchain drivers we should avoid epoch issues. Have you updated to blockchain driver?
UPDATE: So I swapped the hard drives around the rigs and now the other rig is running with 19MH instead... I'm pretty sure all my settings are exactly the same... Is there something windows could have changed after a few too many crashes?
Just clone the good one and forget :-) Surely Windows can accumulate errors, and usually it is much easier to reinstall/clone then to find the error.
PS If it is not too much to ask please let me know what actual EPOCH # are used by the miner in each case. I am still trying to understand what's wrong with my case.