SAP America Inc v. Valtrus Innovations Ltd

1. Case Identification

• Case Number: IPR2025-00417
• Patent #: 6,823,409
• Filed: January 14, 2025
• Petitioner(s): SAP America, Inc.
• Patent Owner(s): Valtrus Innovations Ltd. and Key Patent Innovations Ltd.
• Challenged Claims: 1-17

2. Patent Overview

• Title: Petition for Inter Partes Review of U.S. Patent No. 6,823,409
• Brief Description: The ’409 patent describes a mechanism for filtering processor cache snoops in a multi-processor-bus computer system. The invention uses a coherency control module with an active snoop queue (ASQ) to manage memory requests from multiple processor buses and prevent simultaneous access to the same memory address.

3. Grounds for Unpatentability

Ground 1: Claims 1-4, 7-12, and 15-17 are obvious over Khare.

• Prior Art Relied Upon: U.S. Patent No. 6,487,643 ("Khare").
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner argued that Khare, which discloses a multi-node architecture for managing memory requests, teaches every element of the challenged claims. Petitioner asserted that Khare’s "switching agent 140" is a "coherency control module" that connects to multiple processor nodes via buses, satisfying the core architecture of independent claims 1 and 10. Key components of Khare’s switching agent were mapped to claim limitations: the "request manager 141" was identified as the claimed "request module"; the "pending requests buffers" (145, 146, 147) were identified as the "active snoop queue (ASQ) module" for storing and ordering requests; and the "memory manager 149" with its associated "table 143" was identified as the "static RAM interface module" configured to access an address look-up table. Khare’s system was also argued to teach RAM comprising both static portions (processor caches) and dynamic portions (main memory).
  • Key Aspects: This ground asserted that a single prior art reference discloses all features of the claims, functioning as an anticipation-style obviousness challenge. The argument hinged on mapping the functional modules within Khare’s "switching agent" to the specific modules recited in the ’409 patent’s claims.

Ground 2: Claims 5-6 and 13-14 are obvious over Khare in view of Kirkman.

• Prior Art Relied Upon: Khare (Patent 6,487,643) and U.S. Patent No. 6,581,063 ("Kirkman").
• Core Argument for this Ground:
  • Prior Art Mapping: This ground built upon the system disclosed in Khare, arguing it would have been obvious to implement Khare’s "list structures" (the pending request buffers) using the data structures taught by Kirkman. Specifically, Kirkman was cited for its teaching of efficient, two-dimensional, doubly-linked lists with head and tail pointers (for claim 5) and dependency links (for claim 6). Petitioner contended that Khare’s system, which requires managing and removing entries from request lists, would be improved by implementing Kirkman’s well-known and predictable data structures.
  • Motivation to Combine: Petitioner argued a person of ordinary skill in the art (POSITA), when implementing Khare’s system (which could be done in software), would be motivated to use a known, efficient data structure to manage the request lists. Kirkman taught that doubly-linked lists were a common and ubiquitous structure for such purposes, offering benefits like efficient insertion and removal of entries compared to singly-linked lists. Both Khare and Kirkman were presented as analogous art, as both address the efficient processing of data structures and lists in computer systems to avoid performance issues like livelock.
  • Expectation of Success: A POSITA would have a reasonable expectation of success in combining the references. Implementing Khare's list management with Kirkman's doubly-linked lists was a straightforward application of a well-known software design pattern to improve a known system. Petitioner asserted that Kirkman’s disclosure of intersecting linked lists that create an orthogonal, two-dimensional form would predictably integrate with the cross-list dependency management required by Khare’s system.

4. Arguments Regarding Discretionary Denial

• §325(d) - Same or Substantially the Same Art: Petitioner argued denial under §325(d) is unwarranted because the primary references, Khare and Kirkman, were not cited or considered by the Examiner during the original prosecution of the ’409 patent. Therefore, the grounds presented are not cumulative of art previously reviewed by the USPTO.
• §314(a) - Fintiv Factors: Petitioner argued that the Fintiv factors weigh against discretionary denial.
  • Factor 1 (Stay): Neutral, as Petitioner had not yet moved to stay the parallel district court litigation.
  • Factor 2 (Proximity to Trial): Neutral or weighs against denial. Petitioner argued the trial schedule in the parallel Valtrus I litigation is unreliable due to a pending motion to dismiss for lack of standing and the filing of a second, duplicative lawsuit (Valtrus II) which has no set trial date.
  • Factor 3 (Investment): Weighs against denial because the parallel litigation is in its early stages with minimal investment. Further, several challenged claims in the IPR (3, 6, 11, and 14) are not at issue in the litigation.
  • Factor 4 (Overlap): Weighs against denial. Petitioner stipulated that if the IPR is instituted, it will not pursue the same invalidity grounds in the district court. Additionally, the Kirkman reference is not part of the district court contentions.
  • Factor 6 (Other Circumstances): Weighs against denial due to the compelling merits of the petition, highlighting that Khare and Kirkman teach a nearly identical architecture to solve the same problem as the ’409 patent.

5. Relief Requested

• Petitioner requests the institution of an inter partes review and the cancellation of claims 1-17 of the ’409 patent as unpatentable under 35 U.S.C. §103.