Cuda 4.0 Zero Copy

Should a GTX480 work with the Cuda 4.0 Zero Copy stuff (arbitrary post-alloc pinning)?

I tried the simpleZeroCopy example (changing pin_generic_memory to true at the top to enable the new Zcopy) but it failed on the first cudaHostRegister (it passes the feature-check code). Other 4.0 features seem to work (new/delete).

These are the relevant lines from deviceQuery:
Device 0: “GeForce GTX 480”
CUDA Driver Version: 4.0
CUDA Runtime Version: 4.0
CUDA Capability Major/Minor version number: 2.0
Support host page-locked memory mapping: Yes