The specs wouldn't be any different than they would be for a good quality board for TQFP with a pitch in the order of 0.5mm. If you need to support DDR3 (as the A10 does) and high speeds you need a good board anyway in order to make sure that the transmission lines remain true to design parameters. In other words, signal integrity is a concern with these designs and you just can't use a dirt-cheap board. It doesn't matter that the A10 is hand-solderable. You still need to attend to signal and power integrity and that means that a certain class of board is required.

Having done boards with FPGA's as large as 1152 balls (definitely not soldered by hand) I can tell you that good board houses produce good product and the difference wouldn't be much, if any between a controlled-impedance board for an A10 project versus an equivalent-sized BGA device.

This is important, so I'll repeat it: The A10 requires a controlled-impedance board.

EDIT: Just learned that the A10 requires DDR3. So BGA's are the order or the day. Hand-soldering with a soldering iron is simply not a reality for such a board. If you want to assemble one yourself you have to use a stencil and an oven or hot plate for reflow. Not that hard for small quantities.

I should also note that I've only done this when absolutely necessary. It makes a lot more sense to pay to have SMT boards assembled. It's not that expensive and assembly shops have all of the right equipment to get it right. With BGA's you really need to have full inspection (including X-Ray) in order to ensure proper attachment. This is particularly true in the case of high-speed boards.