This article shows how the architecturalmod- elization of biological retina allows real-time performances on cheap and widespread computing systems. Firstly we describe the biological retina in terms of its pipeline ar- chitecture detailing its layer behaviors and properties. Then we propose a corresponding pipelined model of ar- tificial retina based on cellular automata. In this work, the main innovation is the computing method based on the programming of a personal computer graphical card using OpenGL shading language. The last section demon- strates efficiency of our model with numerical and graph- ical results. We particularly highlight that our direct im- plementation on the Graphical Processor Unit (GPU) provides a computation power about twenty times as fast as ordinary programming.