PTAB

IPR2020-01608

NVIDIA Corp v. Advanced Cluster Systems Inc

Log In
Key Events
Petition
petition Intelligence

1. Case Identification

2. Patent Overview

  • Title: Computer Cluster System and Method
  • Brief Description: The ’289 patent describes a system for parallelizing functions from computer algebra programs (e.g., Mathematica) across multiple processors in a computer cluster. The invention purports to improve upon prior art master-slave architectures by implementing a peer-to-peer relationship that allows direct intercommunication among all kernel modules in the cluster.

3. Grounds for Unpatentability

Ground 1: Obviousness over the Distributed Maple Publications - Claims 1, 4-6, 8, 10-11, 13, 16-19, 21-23, 27, and 29-32 are obvious over Schreiner1 in view of 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 (source code), and the Maple Guide (a 1998 software guide). These collectively describe the "Distributed Maple" system.
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that the Distributed Maple system, a single, cohesive project described across the cited references, discloses every limitation of the challenged claims. The system was designed for parallel computer algebra on networked clusters and explicitly implemented a peer-to-peer communication architecture where all nodes could communicate directly, contradicting the ’289 patent’s assertion that this was novel. The Distributed Maple system utilized "Maple kernels" (analogous to the claimed "kernel modules") as computational engines on each node and a Java-based coordination layer (comprising "dist.Scheduler" and "dist.maple" components, analogous to the claimed "cluster node modules") to manage communication and task distribution between kernels.
    • Motivation to Combine: A POSITA would combine the Distributed Maple publications because they all describe different aspects of the same academic project, were written by the same principal author, expressly reference each other, and were available on the same project website. A POSITA would have consulted the Maple Guide because the Distributed Maple system was built upon and required the use of the Maple software, and the primary references directed a reader to do so for more detail.
    • Expectation of Success: A POSITA would have a high expectation of success, as the references do not describe theoretical components but rather document a single, pre-existing, and operational software system.

Ground 2: Obviousness over Ground 1 References, Howard, and Maple Reference - Claim 14 is obvious over the references in Ground 1 in further view of Howard and the Maple Reference.

  • Prior Art Relied Upon: The references from Ground 1, further combined with Howard (Application # 2003/0195938) and the Maple Reference (a 1997 book).
  • Core Argument for this Ground:
    • Prior Art Mapping: This ground specifically addressed claim 14, which adds a call to calculate a Fourier transform in parallel. Petitioner contended that the Schreiner references already taught an "advanced functions module" for parallelizing various mathematical operations. Howard taught a specific method for calculating a Fourier transform in parallel across a cluster by distributing "slices" of the data to different nodes for independent processing. The Maple Reference confirmed that a standard Fast Fourier Transform (FFT) function was a known part of the Maple software library.
    • Motivation to Combine: A POSITA would combine Howard’s parallel processing technique with Schreiner's Distributed Maple system to achieve the well-known goal of speeding up a common and computationally intensive mathematical operation. Schreiner expressly motivated parallelizing any function to improve performance. Adding a parallel FFT to Schreiner’s existing toolkit of parallel functions would have been a predictable and desirable improvement.
    • Expectation of Success: A POSITA would have a high expectation of success because Howard provided a clear, step-by-step process for parallelizing the FFT, and Schreiner provided a robust framework for distributing computational tasks across a cluster. Integrating Howard's known parallelization algorithm into Schreiner's existing parallelization framework would have been a straightforward application of known techniques.

4. Arguments Regarding Discretionary Denial

  • Petitioner argued that discretionary denial under Fintiv was inappropriate. The core arguments were that: (1) the parallel district court proceeding was at a very early stage, with no claim construction or expert discovery completed; (2) the court-scheduled trial date was set for two to three months after the Final Written Decision (FWD) deadline, with a high likelihood of being further delayed; and (3) Petitioner stipulated that it would not pursue in the district court the specific invalidity grounds raised in the IPR petition, thus minimizing any overlap of issues between the two forums.

5. Relief Requested

  • Petitioner requests the institution of an inter partes review and the cancellation of claims 1, 4-6, 8, 10-11, 13-14, 16-19, 21-23, 27, and 29-32 of the ’289 patent as unpatentable.