PTAB

IPR2021-00178

Google LLC v. Singular Computing LLC

Log In
Key Events
Petition
petition

1. Case Identification

2. Patent Overview

  • Title: Low Precision High Dynamic Range Processing
  • Brief Description: The ’273 patent relates to a device comprising at least one “low precision high dynamic range” (LPHDR) execution unit. The invention is functionally defined by the execution unit’s ability to perform an operation (e.g., multiplication) with a specified minimum level of imprecision on inputs spanning a high dynamic range.

3. Grounds for Unpatentability

Ground 1: Claims 1-70 are obvious over Bates-2010

  • Prior Art Relied Upon: Bates-2010 (Application # 2010/0325186).
  • Core Argument for this Ground: Petitioner’s primary argument is that the challenged claims are not entitled to the 2010 priority date of the parent ’201 Application because its disclosure fails to meet the written description and enablement requirements of 35 U.S.C. §112. Consequently, the published version of that application, Bates-2010, qualifies as prior art under 35 U.S.C. §102(b) and renders the challenged claims obvious under §103.
    • Prior Art Mapping: Petitioner argued that Bates-2010 discloses all limitations of the challenged claims. For independent claim 1, Bates-2010 discloses a device (e.g., a SIMD processor) comprising LPHDR execution units adapted to perform operations. It describes a floating-point embodiment that operates on a numerical range wider than claimed (“one millionth up to one million”). Crucially, while not explicitly linking a specific structure to a specific performance, Bates-2010 suggests the desirability of achieving the claimed functional performance characteristics—namely, that for at least 5% of inputs (the “X” value), the output differs by at least 0.05% (the “Y” value) from an exact calculation. Petitioner contended that these disclosures, combined with the knowledge of a Person of Ordinary Skill in the Art (POSA), render the claims obvious.
    • Motivation to Combine (or Modify): The petition asserted that a POSA would have been motivated to implement the specific performance characteristics suggested as desirable in Bates-2010. Bates-2010 discloses the components (a floating-point unit) and separately suggests the target error metrics. A POSA would have understood that achieving these metrics was a matter of routine design optimization. For example, a POSA would know to reduce the number of mantissa bits in the disclosed floating-point unit to intentionally introduce imprecision, thereby achieving the claimed error rates in exchange for benefits like reduced power consumption or increased speed.
    • Expectation of Success: A POSA would have had a reasonable expectation of success in achieving the claimed performance. The relationship between the number of bits used to represent a number (particularly the mantissa in a floating-point system) and the resulting precision of calculations was well-understood and predictable. Petitioner argued a POSA could use conventional software emulation to determine the exact number of bits required to produce the target error rate before implementing the design in hardware.

4. Key Technical Contentions (Beyond Claim Construction)

  • Failure of Priority Claim: The central technical contention underpinning the entire petition is that the ’273 patent is not entitled to the priority date of its parent ’201 Application. This argument rests on two key failures under §112:
    • Lack of Written Description: The challenged claims cover a broad, functionally defined genus of "execution units" encompassing any implementation technology (e.g., silicon-based digital, analog, nanomechanical, DNA computing) that meets the claimed performance metrics. However, the ’201 Application’s specification allegedly fails to demonstrate that the inventor possessed this broad genus. It describes only one species (a conventional silicon transistor-based digital implementation) in any detail and fails to describe a representative number of species or common structural features that would support the full scope of the genus, particularly regarding the claimed nascent and unpredictable technologies.
    • Lack of Enablement: Petitioner argued the ’201 Application does not enable a POSA to practice the full scope of the claimed invention without undue experimentation. This is particularly true for the claimed embodiments using analog representations or speculative, nascent technologies like DNA or nanomechanical computing, for which the specification provides no guidance or working examples.

5. Arguments Regarding Discretionary Denial

  • Arguments Against Fintiv Denial: Petitioner argued that discretionary denial under §314(a) would be inappropriate because the parallel district court litigation (Singular Computing LLC v. Google LLC, D. Mass.) was in its earliest stages.
    • Trial Timing & Investment: No trial date had been set, and the court acknowledged it was "very far away from trial," with the global pandemic likely causing further delays. Litigation investment was minimal, as the petition was filed diligently after infringement contentions were served, and fact discovery was in its infancy.
    • Overlap of Issues: While some overlap exists, the IPR addresses 69 claims not asserted in the litigation, meaning the Board’s review would resolve issues that the district court would not.
    • Merits: Petitioner asserted that the strong merits of the petition, demonstrating the claims are unpatentable, weigh heavily against denial.
  • Arguments Against §325(d) Denial: Petitioner contended that denial under §325(d) was improper because the petition raises art and arguments not previously considered by the USPTO. The core of the petition—the challenge to the priority claim that makes Bates-2010 available as prior art—was never before the examiner during prosecution.

6. Relief Requested

  • Petitioner requested the institution of an inter partes review and the cancellation of claims 1-70 of Patent 8,407,273 as unpatentable.