Nothing special?

I mean, inference engine might need to get some tweaks, to support whatever compute is available. But then, if you put a few terabytes of disk for swap, and replace RAM to bigger sticks if possible, it should work? Slowly, of course, but there is no reason it should not to.