I'm curious how you arrived at this - or the best diff setting for any miner.  Since it's a statistical process, you can't have two running and compare them directly.  And on H2C the results are all combined in your account, so you cannot see the results for each miner.  The reported hashrate varies all over the map for each proposed nonce.

You'd need to have something like 100 running at each difficulty level and using the same network link (because speed and latency affects hashrate).  Are comparison stats posted somewhere?  It would be cool to find a source of actual data.