Sun Microsystems, Inc.

Oracle America, Inc.

Mark E Pascual

Michael G Lavelle

Nandini Ramani

Patrick Shehane

Ewa M Kubalska

Michael Wasserman

Wing-Cheong Tang

Brian D Emberling

David C Kehlet

Michael Frank Deering

Nathaniel David Naegle

Daniel Chao

Assana Fard

Paul R Ramsey

Kevin Tang

Michael W Schimpf

Yi-Ming Tian

A multi-chip system and method are disclosed that utilizes a plurality of graphics pipelines to perform large kernel convolution. Each graphics pipeline includes a standard rendering unit and a video data convolve unit. Each video data convolve unit receives video pixel data from the video output of the standard rendering unit. The video data convolve units are connected in a chain. Each group of one or more video data convolve units in the chain convolves the video pixel data received by the group. The last video data convolve unit in the chain outputs a stream of convolved pixels.

Large-kernel convolution using multiple industry-standard graphics accelerators

A rendering unit positions a supertile so that it intersects a primitive. The rendering unit repeatedly walks over bins of the supertile, applying a layer of texture to the bins of the supertile in each iteration of said repeated walking. The rendering unit advances to the next texture layer after having applied the current texture layer to each candidate bin of the supertile. The results of each texture layer application to the bins may be stored in a texture accumulation buffer. The size of the supertile corresponds to the size of the texture accumulation buffer. After applying a last layer of texture to the bins of the supertile, the supertile may be advanced to a new position. The rendering unit traverses the primitive with the supertile so that the union of areas visited by the supertile covers the primitive.

Multi-texturing by walking an appropriately-sized supertile over a primitive

A graphics system and method for rendering a plurality of triangles. Information regarding the triangle may first be received. The method may then determine the longest edge or major edge of the triangle and also determine the direction or axis of the longest edge of the triangle. The method may then perform edge walking on the major edge (e.g., along the axis of the major edge) of the triangle, followed by span walking. The edge walking is preferably always performed on the major or longest edge of the triangle, prior to the span walking, and regardless of the orientation of the major edge of the triangle. This operates to load balance the edge walker and the span walker for the plurality of triangles.

Method for reduction of possible renderable graphics primitive shapes for rasterization

A system and method for assigning operations to multiple pipelines in a graphics system is disclosed. The graphics system may include an arbitration unit coupled to a plurality of calculation pipelines. The arbitration unit is operable to provide graphics operations to selected ones of the calculation pipelines. Each of the calculation pipelines is operable to perform a graphics operation. Each of the calculation pipelines may include digital logic and/or a processing element for performing the graphics operations. An operation may be assigned to a pipeline if the pipeline is performing a low latency operation. A low latency operation may comprise an operation that is performed by one of the calculation pipelines in less time than a pre-determined number of clock cycles.

Arbitration scheme for efficient parallel processing

A system and method are disclosed for utilizing a Z slope test to select polygons that may be candidates for multiple storage methods. The method may calculate the absolute Z slope from vertex data and compare the calculated value with a specified threshold value. In some embodiments, for polygons that have an absolute Z slope less than the threshold value, parameter values may be rendered for only one sample position of multiple neighboring sample positions. The parameter values rendered for the one sample position may then be stored in multiple memory locations that correspond to the multiple neighboring sample positions. In some embodiments, storing parameter values in multiple memory locations may be achieved in a single write transaction. In some embodiments, utilization of the Z slope test method may be subject to user input and in other embodiments may be a dynamic decision controlled by the graphics system.

Z-slope test to optimize sample throughput

A system and method for rasterizing and rendering graphics data is disclosed. Vertices may be grouped to form primitives such as triangles, which are rasterized using two-dimensional arrays of samples bins. To overcome fragmentation problems, the system's sample evaluation hardware may be configured to over-evaluate samples each clock cycle. Since a number of the samples will typically not survive evaluation because they will be outside the primitive being rendered, the remaining surviving samples may be combined into sets, with one set being forwarded to subsequent pipeline stages each clock cycle in order to attempt to keep the pipeline utilization high.

Over-evaluating samples during rasterization for improved datapath utilization

A graphics system and method are disclosed that may optimize the rate of pixel generation to match the rate at which a memory may be designed to receive pixel data. If a memory is configured to store multiple pixels substantially simultaneously, it may be advantageous to render an equivalent number of pixels substantially simultaneously and at the same rate. An edge walker that utilizes multiple sets of accumulators to generate multiple scan lines substantially simultaneously and a span walker that utilizes multiple sets of accumulators to render multiple pixel values substantially simultaneously is described.

Method for rasterizing graphics for optimal tiling performance

A system and method for rendering a polygon, such as a triangle. The method may comprise receiving geometry data (or vertex data) defining vertices of the polygon. The method may compute initial vertex x,y values at end points proximate to each of the vertices of the polygon, and a slope value along each edge of the polygon. The computed slope may be a quantized slope value having a first number of bits of precision. The first number of bits of precision may produce inaccuracies for interpolated x,y values computed at the end points of an edge of the polygon. The method may then interpolate x,y values along each respective edge of the polygon using the computed slope along the respective edge of the polygon. Finally the method may store final x,y values for each respective edge of the polygon. The final x,y values comprise the interpolated x,y values for non-end points of the respective edge, and the computed initial vertex x,y values for each of the end points of the respective edge. The operation of storing the computed initial vertex x,y values for each of the end points of the respective edge, instead of using interpolated x,y values at the end points, operates to prevent inclusion of an extraneous pixel and/or exclusion of a pixel within the polygon.

End point value correction when traversing an edge using a quantized slope value

A graphics system configured to operate on a collection of vertices to determine mappings from an initial order to secondary and tertiary ordering. The initial order corresponds to the ordering of the vertices in an input buffer. The secondary (tertiary) ordering corresponds to the ordering of the vertices along a triangle major (minor) axis. The graphics system computes horizontal and vertical displacements along edges of the triangle in the initial ordering, and uses the signs of the horizontal displacements and vertical displacements to access a mapping table which determines the mappings. The mappings may be used to rasterize the triangle in terms of pixels (or samples).

Mapping time-sorted to direction-sorted triangle vertices

A graphics system may be configured to render anti-aliased dots in terms of samples and to generate pixels by filtering the samples. The pixels are supplied to one or more display devices. The means used to generate the samples may perform the computation of radial distance at positions on a grid in a rendering coordinate space, and interpolate estimates for the radial distances of samples around the dot as needed based on the radii at the grid positions.

Growing community of inventors

San Jose, CA, United States of America

Mark E Pascual