Quote from: Vlad2Vlad on August 18, 2013, 09:45:38 AM

Quote from: TwinTurbo on August 18, 2013, 07:57:53 AM

I would love for this new card to be real, but I have a hard time believing this is achievable. I hope someone can prove me wrong. Let's look at this from a competition perspective:

Avalon: 110nm, 67 GH/s, 600+ watts
BFL: 65nm, 60 GH/s, 300+ watts
KNC Miner: 28nm, 400+ GH/s, ~1,000 watts (Yet to be seen)

BFL Monarch: 28nm, 600 GH/s, 350 watts

So the die size cutting nearly in half from the Avalon's 110nm to the BFL's 65nm cut the power in half for roughly the same performance (60-67 GH/s). Avalon is 600+ watts, BFL single is 300+ watts. Makes sense.

One would assume that cutting the die size in half again to 28nm from 65nm would allow the power to be cut in half again for the same 60 GH/s performance. Let's assume it consumes 150 watts as a result (half that of a Single). Seems reasonable.

But now let's assume instead of 60GH/s, the performance is going to be 600 GH/s. If we are going to increase the hash rate by 10x, then the power requirement will also likely increase by 10x. That means the power should be roughly 1,500 watts (about the maximum you can safely draw from a single residential power circuit in the US).

If memory serves me correctly, KNC miner estimated 1,000 watts for their rig. If the BFL Monarch is going to perform 50% greater and the same die size, then power draw naturally would also seem to be 50% greater. So, that seems consistent with my 1,500 watt figure above.

It would be great to get some clarification from BFL on this before placing orders.

Thoughts anyone?

BFL could be fudging their numbers (again) to get orders? Nah, not possible.

Spot on with the fudge comment!