PTAB

IPR2015-01070

Samsung Electronics Co Ltd v. NVIDIA Corp

Key Events

Petition

petition Intelligence

Title: System, method and article of manufacture for shadow mapping
Brief Description: The ’372 patent discloses a method for 3D graphics processing, specifically an improved shadowing algorithm. The invention purports to increase flexibility by "decoupling" lighting variables (e.g., ambient, diffuse, specular) into two separate, sequential shading calculations within a graphics pipeline, rather than combining them in a single function.

Prior Art Relied Upon: Gossett (Patent 6,236,413).
Core Argument for this Ground:
- Prior Art Mapping: Petitioner argued that Gossett discloses a graphics processing system with all elements required by the claims. Gossett’s "texture environment unit" performs multi-pass shading calculations using a "recirculation pipe" that functions as a feedback loop. This system explicitly performs two separate shading passes to implement lighting: a first pass combines texture color with interpolated ambient and diffuse color variables, and a second pass adds a separate specular color variable to the result of the first pass. This two-step process inherently uses "decoupled variables" as claimed, with ambient/diffuse variables used in the first calculation and the specular variable used in the second. The dependent claims are met as Gossett discloses the use of ambient, diffuse, and specular color variables.

Prior Art Relied Upon: Gossett (Patent 6,236,413) and McCool (“Texture Shaders,” HWWS ‘99 Proceedings of the ACM SIGGRAPH/EUROGRAPHICS workshop on graphics hardware, Aug. 1999).
Core Argument for this Ground:
- Prior Art Mapping: Petitioner asserted that Gossett provides the primary system architecture, including a shading module (texture environment unit), a feedback loop (recirculation pipe), and a combiner module (light environment unit). McCool, which is cited by Gossett, explicitly teaches a key missing element: a texture look-up module that retrieves texture information using texture coordinates generated from the output of the shading module itself. McCool explains that to support shaders requiring multiple texture lookups, "the output of a shader needs to be fed back to the texture lookup unit(s)."
- Motivation to Combine: A POSITA would combine McCool's flexible, multi-pass texture lookup functionality with Gossett's graphics processing system to improve realism and performance. Both references address multi-pass shading in graphics pipelines, and the art was moving toward more aggressive and programmable per-pixel shading. Implementing McCool's advanced texturing techniques in a system like Gossett's was a predictable design choice to achieve these known industry goals.

Prior Art Relied Upon: McCool (“Texture Shaders,” HWWS ‘99 Proceedings).
Core Argument for this Ground:
- Prior Art Mapping: Petitioner contended that McCool's proposed extensions to the OpenGL API for programmable shading directly anticipate the claimed method and system. McCool provides an example shader program that performs multiple, sequential shading calculations. A first calculation (ShaderModulate()) multiplies texture variables a and b. A subsequent calculation (ShaderAdd()) adds the result of the first calculation to the product of two different texture variables, c and d. This sequence constitutes a first shading calculation whose output is used in a second shading calculation with a plurality of "decoupled variables" (a and b used in the first; c and d used in the second), thus anticipating all limitations of claim 1. McCool also discloses the system components to implement this, including a texture shader, texture lookup units, and a fragment operations unit for combining outputs.

Prior Art Relied Upon: McCool (“Texture Shaders,” HWWS ‘99 Proceedings) and Ackermann (“An Architecture for a High Performance Rendering Engine,” 1993).
Core Argument for this Ground:
- Prior Art Mapping: McCool discloses the fundamental programmable shading system, including the use of pbuffers to hold intermediate results for multi-pass algorithms. Ackermann explicitly teaches the implementation of a hardware "feedback loop" to support the very type of iterative shading algorithms needed for "shade-trees," a concept also discussed in McCool. Ackermann states, "To implement shade-trees, iterative shading algorithms must be supported... We want to implement such a feedback loop for each of the two internal stages of the pixel modification stage."
- Motivation to Combine: A POSITA seeking to implement the iterative "shade-tree" shading discussed in McCool would have been motivated to consult Ackermann for a concrete hardware solution. Ackermann teaches that an explicit feedback loop is the proper way to implement the iterative algorithms required for shade-trees. Therefore, it would have been obvious to incorporate Ackermann's explicit feedback loop into McCool's system to fully enable the advanced shading techniques McCool describes.

Petitioner argued for constructions based on the Patent Owner's own contentions in a related ITC litigation, asserting they represented the Broadest Reasonable Interpretation.
"shading calculation": "a computation of a value concerning the appearance of a surface."
"decoupled variables": "variables used in two shading calculations where at least one of those variables is used in the first shading calculation but not the second, and at least one of the other variables is used in the second shading calculation but not the first."
"combiner module": "circuitry for combining the output generated by the shading module."

Petitioner requests institution of an inter partes review and cancellation of claims 1-6, 9-16, and 19-25 of the ’372 patent as unpatentable.