PTAB

IPR2025-00243

MediaTek Inc v. Daedalus Prime LLC

Log In
Key Events
Petition
petition

1. Case Identification

2. Patent Overview

  • Title: Power Balancing for Multicore Processors
  • Brief Description: The ’316 patent discloses a power balancing mechanism for dynamically allocating a total power budget among different domains of a multicore processor, such as a processor core domain and a graphics engine domain, to manage performance and power consumption.

3. Grounds for Unpatentability

Ground 1: Obviousness over Felter - Claims 8 and 12 are obvious over Felter.

  • Prior Art Relied Upon: Felter (Application # 2006/0288241).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued Felter discloses all limitations of independent claim 8. Felter teaches a "power administrator" (power controller) in a multi-component system that determines a total power budget for a time interval and allocates it among processor components (domains). The allocation method involves assigning each component its required "standby power" (minimum reservation value) and then proportionally distributing the remaining "available power" based on each component's predicted active power needs, which corresponds to a "sharing policy value." Felter further discloses controlling component frequency via voltage-frequency scaling and throttling based on the allocated power.
    • Motivation to Combine (for §103 grounds): Not applicable as this is a single-reference ground. For dependent claim 12, Petitioner asserted that after Felter's system throttles (denies a request for higher frequency) a processor core due to power limits, a person of ordinary skill in the art (POSITA) would find it obvious to increase that core's predicted utilization for the next time interval. This would, in turn, increment its sharing policy value to allocate more power, a predictable modification to improve performance based on recent demand.
    • Expectation of Success: A POSITA would expect success in modifying the sharing policy based on throttling events, as Felter teaches that "very recent activity is a good predictor of current activity," making this a logical feedback mechanism to improve system responsiveness.

Ground 2: Obviousness over Finkelstein - Claims 13, 14, and 17 are obvious over Finkelstein.

  • Prior Art Relied Upon: Finkelstein (Application # 2010/0115304).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner contended Finkelstein discloses the system of independent claim 13. Finkelstein describes a multicore processor system with multiple "power planes" (domains), including a first domain for processor cores, a second for a graphics controller, and a third containing "power management logic" (system agent circuitry). This logic determines and dynamically allocates a variable power budget between the processor and graphics domains based on "user (or application defined) preferences," which function as power sharing values. Petitioner argued it would be obvious for the power management logic to operate on a fixed budget to reliably control the variable-workload domains.
    • Motivation to Combine (for §103 grounds): Not applicable as this is a single-reference ground. For dependent claim 14, Petitioner asserted Finkelstein teaches that a "single power plane may obtain the entire budget," making it obvious to allocate substantially all power to the processor core domain for a processor-heavy workload and then, for a subsequent graphics-heavy workload, allocate substantially all power to the graphics domain to optimize performance.
    • Expectation of Success: A POSITA would expect success in alternating power allocation between domains based on workload, as this was a known technique for managing performance in systems with both CPU and GPU components.

Ground 3: Obviousness over Finkelstein and Felter - Claims 8 and 12-17 are obvious over Finkelstein in view of Felter.

  • Prior Art Relied Upon: Finkelstein (Application # 2010/0115304) and Felter (Application # 2006/0288241).

  • Core Argument for this Ground:

    • Prior Art Mapping: This combination asserted that Finkelstein's multicore processor architecture, when combined with Felter's specific power allocation method, renders the challenged claims obvious. The argument is that a POSITA would modify Finkelstein’s general pro-rata allocation system to incorporate Felter’s "modified pro-rata technique." This technique explicitly allocates a minimum reservation value to each domain first before distributing the remaining power budget, directly teaching the limitations of claims 8 and 16.
    • Motivation to Combine: A POSITA would combine Finkelstein and Felter to enhance Finkelstein's system with a more robust power allocation scheme. Specifically, a POSITA would incorporate Felter’s method of guaranteeing a minimum power reservation to ensure that all components in Finkelstein’s system have enough power for "minimal work," thereby improving system stability and preventing component failure, especially for critical applications.
    • Expectation of Success: A POSITA would have a reasonable expectation of success because both references address power management in similar multicore processor environments. Integrating Felter's specific allocation formula into Finkelstein's more general framework would be a predictable and straightforward design choice to achieve the benefits of both systems.

  • Additional Grounds: Petitioner asserted additional obviousness challenges (Grounds 4-6) based on combinations including Therien (Application # 2010/0162006). These grounds argued Therien explicitly teaches using an "increase request flag" and "preference values" to modify power budgets, further supporting the obviousness of claims related to incrementing sharing values (claims 12, 15) and favoring one domain over another (claim 17).

4. Arguments Regarding Discretionary Denial

  • Petitioner argued discretionary denial under Fintiv is not warranted. Key reasons included that Petitioner was diligent in filing the petition, the parallel district court proceeding is in its early stages with minimal investment, final contentions have not been exchanged, and the petition challenges claims not asserted in the litigation. Petitioner contended the compelling merits of the petition also weigh heavily in favor of institution.

5. Relief Requested

  • Petitioner requests institution of inter partes review (IPR) and cancellation of claims 8 and 12-17 of the ’316 patent as unpatentable.