How to call atomidAdd() in CUDA kernel

My GPU card is notepad Geforce GT 525M, with computation capbility 2.1, but when I call atomicAdd() in CUDA kernel, it shows the identifer of “atomicAdd()” is not defined. I do not know why, does anyone can help me work this problem out, thanks very much.

Try to compile your program with " -arch=sm_20 ".

It works. Thanks very much. Can I ask your question again, does " -arch=sm_20 " mean computation capbility 2.0, and so on?