A system is disclosed for improving the flexibility and performance of video walls including a method for using a primary GPU for initial rendering to a GPU frame buffer, copying of this frame buffer to system memory for processing into multiple sub-frames then outputting the sub-frames via multiple secondary graphics controllers. This system enables the video wall server to leverage performance advantages afforded by GPU acceleration and maintaining performance while providing full flexibility of the CPU and system memory to apply the required transformations to the sub-images as well as flexibility in the selection of secondary graphics controllers (including network graphics approaches where the graphics controller is connected over a network) for outputting the multiple sub-images to a plurality of displays. This has applications generally in the field of real-time multiple display graphics processing as well as specific applications in the field of video walls and network video walls. A method and computer readable medium are also disclosed that operate in accordance with the system.