It seems illogical that they would design multiple different boards with different FPGA given how small the market is. The "single" would be fine for use in a cluster, maybe having different connectors, and communication methods but roughly the same design. Having a different board makes little sense. I mean 1.05GH using 2 FPGA for one product and then 1.7GH per board for the second product.
If the rig consisted of similar dual FPGA designs one could expect it to be roughly 1.05 * 32 = 33.6GH/s. Even +10% would be 36.96GH which makes the 54.4 number illogical. Not the first thing w/ this company.
In any case I am hoping we'll have a better grasp of the situation in 2 or 3 weeks. Let's see.
Given the high cost of FPGA that would be beyond stupid. It would be like car company selling 8 cylinder cars and then making a compact model which has 4 cylinders by just turning off 4 of the cylinders (but keeping all the cost of 8 cylinder engine) and putting lackluster cooling in it.
FPGA is very very very expensive. Every clock cycle is worth something. You don't build a product which could run at 1.7 GH and say "well 1 GH is good enough because we fucked up the cooling and power requirements." It just doesn't make any sense.