This cant be that complicated of a piece of hardware. I wish there was more demand for them so we could get a alternative made up.
no it really is complicated and expensive. one is a PCI-ex board using an FPGA to do Direct Memory Access Control (DMAC for short hence iDmac)
the other is another fpga to encapsulates the inputs in to a LVDS stream. LVDS is used in a lot of laptops as a display protocol. EG its super fast.
recreating the IO would be the simplest but most likely the most expensive but the iDmac will most likely be the cheapest to build but very difficult to pull off as its not a simple network card or the likes.
DMAC is very easy to fuck up and as you can probably tell when it fucks up it can "REALLY" fuck up.
I know of 2 people in the scene that "could" do it. but its not a weekend job so it would need to be a for profit job
Is there any actual benefit to going "real hardware" in this case?
as atrfate said you can run the FIO on a normal PC. thats actually how most of my dev work with this has been. this is also true for the JVS input card on the TTX and in turn a card reader.
so technically no but i like other prefer to have legit setups so thats how we support it.
as for the lag. due to the way FIO works on the application side it is near on lag free.
It is also less resource intensive as the io polling is not exactly handled by the game.
all the game does is send a command to the dmac and it handles everything and then edits some memory. all away from the cpu
now their IO board itself and the LVDS stream from what i have heard is not the best implementation. So i cant comment really on any lag there.
@android may be able to give a bit more info here.
For FIO and PC games on a TTX2/3 i do plan to do a FIO to vjoy tool for this purpose but right now i have zero time to commit to anything really. this is one of them projects i need to set aside a whole weekend and just go from start to finish on it