the ps3 already has a F@H client, i dont thing the 360 will get one, or it would have by now.
if you have two cards, just run two clients at the same time, i doubt an SLi optemised version will do much apart from make it slightly simpler in that you dont have to set up two clients, one for each card.
I saw a presentation on the algorithms used in F@H a while back (in April). They said they were kicking around ideas for a multi-GPU implementation, but didn’t have any concrete plans at the time. Not sure what they are doing now. I would guess they are hard at work getting functionality into OpenMM.
I hope you misunderstood me. What i am talking about is, for example using 3 GTX 275s in 3 way SLI and scaling 1 F@H client among them. I am an avid folder with about 5 rigs ATM and have all of my systems running 1 client per GPU and 2-3 clients per machine.
In gaming scaling does not exactly split the load 50-50 or 33-33-33. However i am in belief that it can be done with F@H.
In fact the reason to scale is to make it easier to setup clients. however it could also unlock some hidden potential that cannot be currently done with a single client on one card.
EDIT- I see where your misunderstanding is. CONSOLE. When I say console I don’t mean gaming console. I mean folding client. There is the Nvidia systray client and the Stanford Console client. The console client runs in a CMD style console and is much more stable than the systray version.
I have absolutely no idea what the F@H code looks like, but I wouldn’t think it would be too hard for them to scale it across multiple GPU’s in the machine by simply assigning each GPU it’s own work unit (or splitting up a work unit in some way, if possible). Whether they have plans to do that, I don’t know.
It’s not going to unlock any hidden potential that you wouldn’t get from a single card, other than the fact that it will scale nearly linearly with the number of devices in the system.