Great, is this algorithm already coded or is this still to come? Do you know or expect the first mining software to be AMD or Nvidia focused?
We are still hashing out just how it will be put together, I would think actual development will start once holidays are done. I would like to equally offer for both AMD and Nvidia, but really that will be dependent on who picks up the miner development and or adds to their existing miner.
Any idea on what would keep this fpga resistant other than the need to develop a new bistreams to accommodate the algorithm?
The newest CN has some elements that really help slow down FPGAS, part of what is being looked at is incorporating those elements across the whole combination. Not to block them completely but slow them down to GPU levels at which point they become uninterested given the high purchase cost of decent FPGA units. Plus adding in the whole range of variants you will end up having to daisy chain larger numbers of medium priced boards to get it to work.
Is it a perfect solution? No, but will it work significantly better than anything out of the box at the moment? Yes

Sounds good but it seems like the perfect Target for GPU acceleration, the GPU would handle those elements and the fpga would handle the parts it is more efficient at. That's the plan for most algorithms when SQRLS miner launches for their acorns.