Well, you can still do double-floats, combining two 32-bit floats for a greater precision. While that doesn't get you full double precision, it just might be enough. And of course you can extend the same idea to implement quad-floats and so on.

http://andrewthall.org/papers/df64_qf128.pdf