basic kernel failure why?

Accelerated Computing CUDA CUDA Programming and Performance
1

this simple example is not working, im testing 2D grids,with 1D blocks inside. im not sure what does it mean with “Invalid Argument”

Cuda error: Kernel execution failed in file '/home/neoideo/Galaxia/meshEditor/src/myCudaCalls.cu' in line 175 : invalid argument.

and code

//-------------------------------------------------------------------------

//! Detectar Colisiones

//-------------------------------------------------------------------------

void myCudaDetectarColisionesShared(){

	dim3 dimBlock(MYCUDA_BLOCK_SIZE_COLISSION_i);

	dim3 dimGrid((numCaras+MYCUDA_BLOCK_SIZE_COLISSION_j) / MYCUDA_BLOCK_SIZE_COLISSION_j, (numCaras+MYCUDA_BLOCK_SIZE_COLISSION_i)/ dimBlock.x );

	cudaThreadSynchronize();

	kernelDetectarInterseccionesShared<<< dimGrid, dimBlock >>>();

	CUT_CHECK_ERROR("Kernel execution failed");

	cudaThreadSynchronize();

}

#define MYCUDA_BLOCK_SIZE_COLISSION_i 32

#define MYCUDA_BLOCK_SIZE_COLISSION_j 10

__global__ void kernelDetectarInterseccionesShared(){

	int i = blockIdx.y * blockDim.x + threadIdx.x;

	if( i > 10 ){

		int a = i+1;

	}

	return;

}