Another method which I have implemented one time, in a CPLD incidently, is to organize the memory into say 16 bit width and have the video fetch read 16 bits at a time. These then get shifted out to the display and the memory, which is normal speed, can be accessed in a byte manner by the processor. This eliminates the need for double speed memory at the expense of making the video processing circuitry twice as wide. Clearly this solution two memory chipss to achieve 16-bits width but if your making a color type display support then more memory can be useful.

Michael Karas