OpenCL Atomics Is atomic support present


Is atomic support present in OpenCL included with CUDA 2.3 and Driver 190.29 ?
I couldnt see any SDK example that uses it.

When I do this call within my OpenCL kernel I get a CL_OUT_OF_RESOURCES

I use a call like the below pointing to global memory.

atom_add(&binResult[mem_index], 1);


Did you enable atomics extension first?

#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable

My GF8800GT supports cl_khr_global_int32_base_atomics, but not cl_khr_local_int32_base_atomics :( Drivers 197.13(3-15-2010), Windows 2003 x64, CUDA 3.0. Somebody with 2.3 might answer better :-)