error: identifier "atomicAnd" is undefined

Accelerated Computing CUDA CUDA Programming and Performance
#1

Hi, i’m getting this error:

error: identifier “atomicAnd” is undefined

should i include something to use atomicAnd, and atomicOr?

here’s the piece of code:

if (Func(sdata[0] - layer->thresholds[outputNeuron], layer->functionType) > 0){

	atomicOr(&(((unsigned*)(layer->outputNeurons))[outputNeuron / BITS_PER_INT]), 

			(0x80000000>>(outputNeuron % BITS_PER_INT)));

} else {

				

	atomicAnd(&(((unsigned*)(layer->outputNeurons))[outputNeuron / BITS_PER_INT]), 

			(0x7EEEEEEE>>(outputNeuron % BITS_PER_INT)));

}

thank you

#2

you need to add -arch sm_11 to your compilation switches.
(and of course have a compute 1.1 enabled card)

#3

Thank you, that worked.

i didn’t find it in the programming guide.