Nested for loops crash my kernel

Accelerated Computing CUDA CUDA Programming and Performance
1

I’ve been playing around with different ways to solve my discrete convolution and whenever I try to use nested for loops, it automatically crashes. I made a simple kernel below which crashes if the second for loop is present.

int idx = threadIdx.x + (blockIdx.x * blockDim.x);

		if(idx < length)

		{

			for(int j = 0; j < length; j++)

			{

				for(int k = 0; k < length; k++)

				{

					outputSignalArray[k] += 2;

				}

			}

		}

Are nested for loops not allowed? If I am trying to use them, does that mean I not using CUDA correctly to begin with?

2

Nested for loops are certainly allowed. Can you post the full .cu file which produces the behaviour you’re seeing? Saying that it ‘crashes’ doesn’t help much.

3

Nested for loops are certainly allowed. Can you post the full .cu file which produces the behaviour you’re seeing? Saying that it ‘crashes’ doesn’t help much.