A method, system, and computer-readable storage medium are disclosed for performing an operation on single-channel input using a plurality of execution channels of a graphics processing unit (GPU). A single-channel input may be divided into a plurality of sub-inputs, wherein each of the plurality of sub-inputs comprises a portion of the single-channel input. Borders between the sub-inputs may overlap. Program instructions may be executed on the GPU to perform the operation. At least a subset of the program instructions may be executed on each of the plurality of execution channels in parallel for a corresponding one of the plurality of sub-inputs. In one embodiment, the single-channel input may comprise a single-channel input image for an image-processing operation. In one embodiment, the GPU may comprise four execution channels.