Architecture emulation to port libraries

I’m porting cminpack to CUDA. (It is a library that includes Levenberg-Marquardt functions).

Currently the port it’s tested OK for computing capability 1.1 (with float -single precision- and
without pointers to functions).

I would like it to support double-precision and pointers to functions but I have not a battery of physical
CARDS available for testing.

How could I emulate different capabilities (1.3 and 2.0 at least)?

I read a lot stuff around and I’m afraid that the only choice is ocelot.
I’m afraid because I’m using MacOs (leopard) and don’t want to get lost in further installation problems.

Is it any decent way to use gdb-cuda to emulate GPUs?

Any comment is welcome ! Thanks