How to use __dadd_rn ? (And __any and __all ?)


I am trying to benchmark all the instructions available in CUDA and I am very frustrated by this one : __dadd_rn that Visual Studio doesn’t recognize and refuse to compile.
I have joined a picture of the errors that VS returns to me. I am working with a GeForce GTX 275.

Thks :)

did you compile with -arch sm_13?

__dadd_rn() is a CUDA device function, and as such only available in device code. The attached picture does not display for me, but from the description it sounds like you are trying to call __dadd_rn() from host code?

Yeah I have compiled with this flag.

in fact, I have gathered all the instructions in one file called instructions.h and I redefine them with macros.

The compiler recognize well all the instructions except __any, __all & __dadd_rn …