Method and system for presenting three-dimensional computer graphics images using multiple graphics processing units
View Patent ↗The present invention provides a method and system for presenting three-dimensional computer graphics images using multiple graphics processing units. The dimensions of the scene to be rendered are bounded by a rectangular volume decomposed into rectangular subvolumes. Vertices of graphics primitives are compared with subvolume boundaries to determine to which subvolume a graphics primitive should be assigned. A GPU is assigned to each subvolume to render the graphics data that lies within it. A viewing position point is determined and communicated to each GPU. Rendered graphics data from each GPU are ordered based upon the viewing position Outputs of the individual GPUs are combined by blending within an image combiners. Outputs of image combiners can be presented for viewing or further combined in a subsequent stage image combiner.
1. A method for presenting three-dimensional computer graphics images using multiple graphics processing units, comprising the steps of:
(1) allocating three-dimensional computer graphics data to rectangular subvolumes that define a scene such that said allocated three-dimensional computer graphics data corresponds to a portion of the scene that lies within the rectangular subvolumes to which the multiple graphics processing units have been assigned;
(2) rendering, by the multiple graphics processing units, said allocated three-dimensional computer graphics data;
(3) combining said rendered three-dimensional computer graphics data, thereby producing a three-dimensional computer graphics image; and
(4) selecting one of presenting a first set of said combined three-dimensional computer graphics data and combining, at a second stage image combiner, the first set of said combined three-dimensional computer graphics data received directly from a first first stage image combiner with a second set of said combined three-dimensional computer graphics data received directly from a second first stage image combiner.
2. The method of claim 1 , further comprising, before step ( 2 ), the steps of:
(5) determining a viewing position.
3. The method of claim 2 , wherein said combining further comprises the step of:
(6) ordering said rendered three-dimensional computer graphics data based on locations between said determined viewing position and the rectangular subvolumes.
4. The method of claim 1 , wherein said combining further comprises the step of:
(5) blending said rendered three-dimensional computer graphics data.
5. A system for presenting three-dimensional computer graphics images, comprising:
memory configured to store three-dimensional computer graphics data;
at least one graphics processing unit configured to render a portion of the three-dimensional computer graphics data that corresponds to rectangular subvolumes to which said at least one graphics processing unit is assigned;
a bus configured to communicate a viewing position to each of said at least one graphics processing unit; and
at least one image combiner configured to combine the three-dimensional computer graphics data rendered by said at least one graphics processing unit to produce a three-dimensional computer graphics image;
wherein at least two first stage image combiners of said image combiners are configured to be selectively coupled directly to one of an output device and a second stage image combiner of said image combiners.
6. The system of claim 5 , wherein said memory comprises memory cells such that each of said memory cells is accessible by only one of said at least one graphics processing unit.
7. A method for presenting three-dimensional computer graphics images, comprising the steps of:
(1) allocating three-dimensional computer graphics data to rectangular subvolumes that define a scene;
(2) rendering said allocated three-dimensional computer graphics data;
(3) combining a first set of said rendered three-dimensional computer graphics data and a second set of said rendered three-dimensional computer graphics data; and
(4) selecting one of presenting a first set of said combined three-dimensional computer graphics data and combining, at a second stage combiner, the first set of said combined three-dimensional computer graphics data received directly from a first first stage combiner with a second set of said combined three-dimensional computer graphics data received directly from a second first stage combiner.
8. A system for presenting three-dimensional computer graphics images, comprising:
memory configured to store three-dimensional computer graphics data;
graphics processing units configured to render portions of the three-dimensional computer graphics data that correspond to rectangular subvolumes to which said graphics processing units are assigned;
a bus configured to communicate a viewing position to each of said graphics processing units; and
combiners configured to combine the three-dimensional computer graphics data rendered by said graphics processing units to produce a three-dimensional computer graphics image;
wherein at least two first stage combiners of said combiners are configured to be selectively coupled directly to one of an output device and a second stage combiner of said combiners.