Report question

Question: Aligning GPU memory accesses of an image convolution (OpenCL/CUDA) kernel