Parallel pipelined merge engines

An image generator comprising:

• a plurality of rendering nodes;
• a network coupling the rendering nodes such that each node is coupled to the next node and a previous node;
• and
• wherein each rendering node of the plurality of nodes includes:
• (a) a processor for generating graphics commands,
• (b) a rendering engine for rendering image parts based on graphics commands received from the processor, and
• (c) a merge engine coupled to the rendering engine to receive local pixel data and coupled to the next node to send merged pixel data to the next node, wherein each merge engine comprises a first plurality of registers for control, status, and address information and a second pluralitv of registers identical to the first plurality, only the second plurality of registers for each merge engine configured to be selectively aeecssed or modified by other merge engines via the network.

An image generator, comprising:

• one or more workstations;
• an image controller implemented as a distributed program;
• a plurality of rendering engines;
• and
• a plurality of merge engines some or all of which having a data path therebetween, wherein the data path is realized as a daisy-chained bus connection or switch network that can control the order of data traffic flow between the plurality of merge engines, wherein at least one of the plurality of merge engines is designated as an output merge engine for providing a final merged image, wherein each merge engine comprises a first plurality of registers for control, status, and address information and a second plurality of registers identical to the first plurality, only the second plurality of registers for each merge engine configured to be selectively modified by other merge engines via the daisy-chained bus coinection or switch network.

An image generation method comprising:

• instantiating an image controller program that is implemented as a multithreaded program;
• instantiating a user application that interacts with a first thread of the image controller program for providing rendering information;
• invoking an application program interface, the first thread accessing a set of graphic accelerator cards via the application program interface;
• instantiating a set of threads, the application program interface distributing the rendering information by communicating with the set of threads, each thread in the set of threads invoking a local application program interface to drive a respective graphic accelerator for producing an image that is passed to a merge engine corresponding to that thread, wherein a final merge engine in a merge pathway deposits a fully processed image in a buffer from which the fully processed image is retrievable by a display;
• and
• dynamically modifying only one of at least two sets of identical control, status, and address registers within a first merge engine through a second merge engine on the merge pathway.

An image rendering system comprising:

• a plurality of objects to be rendered in an image, the plurality of objects comprising at least first and second objects;
• a plurality of rendering engines for rendering the objects and generating associated object pixel data;
• a plurality of merge engines for merging the first and second rendered objects according to a predetermined set of rules;
• a pixel data bus connecting the plurality of merge engines to each other;
• and
• wherein each merge engine comprises a first plurality of registers for control, status, and address information and a second plurality of registers identical to the first pluralitv, onlv the second plurality of registers for each merge engine configured to be selectively modified by other merge engines via the pixel data bus.