OpenCL compiler optimizations enabled SDK 3.0

Is there any documentation regarding the compiler options enabled with NVIDIA GPU Computing SDK 3.0 for OpenCL

are the following options enabled ?

-cl-relaxed-fast-math (is this same as the fast_math option with CUDA ?)

-cl-no-signed-zeros (== -ftz=true ?)

-Werror

I dont get an error when use these options, but at the same time also there no documentation regarding this ?

Also looking for a su;up of changes of the SDK3.0…

Also looking for a su;up of changes of the SDK3.0…