Abstract: A multiply-accumulate (MAC) circuit including a queue circuit programmable at runtime is described. In one embodiment, the queue circuit includes a main queue that is programmable at runtime and a supplementary queue. In one embodiment, the queue circuit further includes M multiplexers coupled to the main queue and the supplementary queue, where M is an integer greater than or equal to one. In one embodiment, the MAC circuit further includes M multiplier elements coupled to the queue circuit and an accumulator circuit coupled to the M multiplier elements. The M multiplier elements receive multiplicands from the queue circuits and provide results of multiplications to the accumulator circuit. The accumulator circuit maintains a running sum of the results of the M multiplications performed by the M multiplier elements. In one embodiment, the accumulator circuit includes two adder circuits.

Abstract: A method of filtering image data is described. In one embodiment, the method includes storing in line buffers image data corresponding to a plurality of rows of an image; filtering image data on one row of multiple of rows; and filtering image data on another row of the multiple rows without changing the image data stored in the line buffers between filtering image data on the one row and filtering image data on the another row.