NVIDIA Corp v. Advanced Cluster Systems Inc

1. Case Identification

• Case #: IPR2021-00075
• Patent #: 8,140,612
• Filed: October 16, 2020
• Petitioner(s): NVIDIA Corporation
• Patent Owner(s): Advanced Cluster Systems, Inc.
• Challenged Claims: 1, 4-8, 10, 12-13, 15, and 29

2. Patent Overview

• Title: Computer Cluster System and Method
• Brief Description: The ’612 patent describes a computer cluster system designed to parallelize mathematical functions across multiple processors. The system uses peer-to-peer communications between nodes, where each node runs a "single-node kernel module" (like Mathematica) and a "cluster node module" to coordinate tasks and communication.

3. Grounds for Unpatentability

Ground 1: Obviousness over Distributed Maple Publications - Claims 1, 6, 10, 12-13, 15, and 29 are obvious over Schreiner1, Schreiner2, Schreiner3, the Distributed Maple Code, and the Maple Guide.

• Prior Art Relied Upon: Schreiner1 (a 2003 journal article), Schreiner2 (a 2001 user manual), Schreiner3 (a 2002 paper), the Distributed Maple Code (software code published by 2003), and the Maple Guide (a 1998 learning guide).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner argued that the combination of references, which all describe a single academic project called "Distributed Maple," discloses every limitation of the challenged claims. The Distributed Maple system was described as a computer cluster of networked nodes, each running a standard Maple kernel (the claimed "single-node kernel module"). The system's dist.Scheduler and dist.maple software components on each node collectively function as the claimed "cluster node modules," managing peer-to-peer communications to distribute and execute parallel tasks. User instructions were entered via a Maple frontend, interpreted by the cluster node modules, and sent to other nodes for parallel execution, with results returned to the user interface.
  • Motivation to Combine: A POSITA would combine these references because they explicitly relate to the same project, "Distributed Maple," share a common author (Dr. Schreiner), and were publicly available together. Schreiner1 provides a system overview and expressly references the other documents and the code for more detail. The motivation to include the Maple Guide was direct, as the Distributed Maple system was built upon and required the use of the Maple software described in the guide.
  • Expectation of Success: The petition asserted a high expectation of success because the references document a single, pre-existing, and operational system. The combination was not a theoretical proposal but a description of an integrated system that had already been successfully built and tested.

Ground 2: Obviousness over Distributed Maple Publications and Deitel - Claims 4-5 and 7-8 are obvious over the references of Ground 1 in further view of Deitel.

• Prior Art Relied Upon: The references from Ground 1, in further view of Deitel (a 2004 operating systems textbook).
• Core Argument for this Ground:
  • Prior Art Mapping: This ground adds Deitel to the primary combination to address claims reciting integration with an operating system (OS). Claim 4 requires an OS to incorporate a cluster node module; Claim 5 adds an OS-provided programming interface (API); and Claims 7-8 require access to partitioned memory space. Petitioner argued that Deitel taught that moving application-level functions (like the communication and scheduling of Distributed Maple) into the OS, exposing them via APIs, and partitioning memory between the OS and user applications were all conventional, well-known techniques for improving performance, security, and networking capabilities.
  • Motivation to Combine: A POSITA would combine the Distributed Maple system with the OS design principles from Deitel to improve efficiency and portability. Schreiner1 itself suggested that a native C implementation using MPI (a function often supported at the OS level) would be more efficient. Deitel provided the known roadmap for such OS-level integration, making it an obvious modification to enhance the Distributed Maple system.
  • Expectation of Success: Success was expected because integrating functions into an OS and providing APIs were standard programming practices and core skills for a POSITA, not requiring undue experimentation.

Ground 3: Obviousness over Distributed Maple Publications, Deitel, and Chen - Claims 4-5 and 7-8 are obvious over the references of Ground 2 in further view of Chen.

• Prior Art Relied Upon: The references from Ground 2, in further view of Chen (a 2001 paper on OS kernel-integrated fault-tolerant MPI).
• Core Argument for this Ground:
  • Prior Art Mapping: This ground reinforces the argument for OS integration (Claim 4) by adding a specific motivation: enhancing fault tolerance. The Schreiner references show that Distributed Maple implemented fault-tolerance features (e.g., task logging, peer checking). Chen taught that integrating such fault-tolerance mechanisms directly into the OS kernel was more "effective, transparent and extensive" and reduced overhead compared to user-space implementations.
  • Motivation to Combine: A POSITA, aware of the fault-tolerance features in Distributed Maple and the benefits of OS-level implementation described by Chen, would be motivated to move Distributed Maple's fault-tolerance logic into the OS kernel. This would create a more robust, efficient, and streamlined system, which was the explicit goal taught by Chen.
  • Expectation of Success: The combination was presented as a predictable application of known software design principles (moving functions to the kernel for efficiency) to a known system (Distributed Maple) to achieve a predictable benefit (improved fault tolerance).

4. Arguments Regarding Discretionary Denial

• Petitioner argued that discretionary denial under §314(a) and the Fintiv factors would be inappropriate. The petition asserted that: (1) it would promptly seek a stay of the parallel district court litigation; (2) the court's trial date was scheduled for after the statutory deadline for a Final Written Decision (FWD); (3) the parallel litigation was in a very early stage, with no claim construction or substantive discovery completed; and (4) Petitioner stipulated that it would not pursue the same invalidity grounds in the district court if the inter partes review (IPR) was instituted.

5. Relief Requested

• Petitioner requests institution of IPR and cancellation of claims 1, 4-8, 10, 12-13, 15, and 29 of the ’612 patent as unpatentable.