I will remove my negative trust once you send a board to dogie and he confirms your numbers.
HF's document mentions 96x4x2 and 422GH/s at nominal 550MHz. GN on our board provides at least 520 GH/s at 550MHz. The screenshot shows what cgminer shows (with original HF driver) and this is what we are providing. The document by HashFast was composed for their equipment and boards and wasn't updated even after Yoli was released that was 2 steps away from discovering the whole potential of Golden Nonce. Even rev 2.1/2.2 of the boards were able, at certain manipulations with cooling and voltage, to reach results that are not documented by HashFast. That is a fact. This chip has a great potential but not everyone can revise and upgrade the board. Your claim is unjustified.
Your claim is bullshit. Your board can not magically create more hashing engines on the chip. You can claim that your board is more efficient or allows the chip to be clocked higher, but stating that you can produce more hashes per clock than is physically possible is complete bullshit.
Can you explain why your image (linked below for posterity incase it gets deleted from imgur) doesn't conform to the expected relationship between hashrate and WU? The WU total seems to indicate they are actually running around 610GH/s or so, probably with the clock actually set to 800MHz.
