CONVOLUTION LAYER PROCESSOR AND METHOD

The disclosure relates to a convolution layer processor for a neural network accelerator and to a method of operating a convolution layer processor. Example embodiments include a convolution layer processor (500) for a neural network accelerator, comprising: a memory access module (501) configured to access elements of an input feature map comprising a first array of pixels from a memory (508); and a plurality of convolution modules (3140-x), each convolution module configured to receive an element of the input feature map from the memory access module (501) and perform a convolution operation on the received element of the input feature map with a convolution kernel comprising a second array of pixels to provide a corresponding element of an output feature map, the memory access module (501) comprising: a DMA requester (503) configured to request elements of the input feature map from the memory (508) via a data bus (504); a data buffer (502) configured to provide the requested elements to each of the plurality of convolution modules (3140-x); and a pad supervisor module (507) configured to provide to the data buffer (502), for each element requested by the DMA requester (503), padding pixels of the input feature map if the requested element extends beyond a boundary of the input feature map.