Advanced Micro Devices, Inc.Download PDFPatent Trials and Appeals BoardMay 6, 20212020001279 (P.T.A.B. May. 6, 2021) Copy Citation UNITED STATES PATENT AND TRADEMARK OFFICE UNITED STATES DEPARTMENT OF COMMERCE United States Patent and Trademark Office Address: COMMISSIONER FOR PATENTS P.O. Box 1450 Alexandria, Virginia 22313-1450 www.uspto.gov APPLICATION NO. FILING DATE FIRST NAMED INVENTOR ATTORNEY DOCKET NO. CONFIRMATION NO. 14/755,401 06/30/2015 Zhe Wang AMD-130567-US-NP 8168 25310 7590 05/06/2021 Volpe Koenig DEPT. AMD 30 SOUTH 17TH STREET -18TH FLOOR PHILADELPHIA, PA 19103 EXAMINER MERCADO, RAMON A ART UNIT PAPER NUMBER 2132 NOTIFICATION DATE DELIVERY MODE 05/06/2021 ELECTRONIC Please find below and/or attached an Office communication concerning this application or proceeding. The time period for reply, if any, is set in the attached communication. Notice of the Office communication was sent electronically on above-indicated "Notification Date" to the following e-mail address(es): eoffice@volpe-koenig.com PTOL-90A (Rev. 04/07) UNITED STATES PATENT AND TRADEMARK OFFICE ____________ BEFORE THE PATENT TRIAL AND APPEAL BOARD ____________ Ex parte ZHE WANG, SOORAJ PUTHOOR, and BRADFORD M. BECKMANN Appeal 2020-001279 Application 14/755,401 Technology Center 2100 ____________ Before ST. JOHN COURTENAY III, SCOTT E. BAIN, and NORMAN H. BEAMER, Administrative Patent Judges. BEAMER, Administrative Patent Judge. DECISION ON APPEAL Appellant appeals under 35 U.S.C. § 134(a) from the Examiner’s final rejection of claims 1, 4–9, 12–17, 19, and 20.1 Claims 2, 3, 10, 11, and 18 are cancelled. We have jurisdiction over the pending rejected claims under 35 U.S.C. § 6(b). We REVERSE. 1 We use the word “Appellant” to refer to “applicant” as defined in 37 C.F.R. § 1.42. Appellant identifies Advanced Micro Devices, Inc., as the real party in interest. (Appeal Br. 3.) Appeal 2020-001279 Application 14/755,401 2 THE INVENTION Appellant’s disclosed and claimed invention is directed to managing performance of a processor having multiple compute units. (Abstract.) Independent claim 1, reproduced below, is illustrative of the subject matter on appeal: 1. A method for managing performance of a processor having multiple compute units, the method comprising: determining an effective number of the multiple compute units to designate as having priority; and when the effective number is nonzero: designating the effective number of the multiple compute units each as a priority compute unit, allowing each priority compute unit to read from and write to a shared cache and to` allocate into the shared cache, and allowing a compute unit that is not a priority compute unit to read from and write to the shared cache and disallowing the compute unit which is not a priority compute unit to allocate into the shared cache. REJECTIONS The Examiner rejected claims 1, 4, 6, 7, 9, 12, 14, 15, 17, 19, and 20 under 35 U.S.C. § 103 as being unpatentable over the combination of Mekkat, (Mekkat, et al., Managing Shared Last-Level Cache In A Heterogeneous Multicore Processor, National Science Foundation, (2013), hereinafter “Mekkat”), Sim, (Sim et al., US 8,996,818 B2; iss. Mar. 31, Appeal 2020-001279 Application 14/755,401 3 2015, hereinafter “Sim”), and Hady, (Hady et al., US 2006/0112227 A1, pub. May 25, 2006, hereinafter “Hady”). (Final Act. 3–7.)2 The Examiner rejected claims 5 and 13 under 35 U.S.C. § 103 as being unpatentable over the combination of Mekkat, Sim, and Harriman (Harriman et al., US 6,112,265; iss. Aug. 29, 2000, hereinafter “Harriman”). (Final Act. 7–8.) The Examiner rejected claims 8 and 16 under 35 U.S.C. § 103 as being unpatentable over the combination of Mekkat, Sim, and Brent (Brent et al, US 2010/0131955 A1; pub. May 27, 2010, hereinafter “Brent”). (Final Act. 8–9.) ISSUE ON APPEAL Appellant’s arguments present the following dispositive issue:3 Whether the Examiner erred in finding that the combination of Mekkat, Sim, and Hady taught or suggested the independent claim 1 limitation, “determining an effective number of the multiple compute units to designate as having priority,” and the commensurate limitations of independent claims 9 and 17. (Appeal Br. 5–8.) 2 The Examiner also indicated that claims 2, 3, 10, and 11 were rejected on this ground, but that appears to be an inadvertent error, because those claims are cancelled. 3 Rather than reiterate the arguments of Appellant and the positions of the Examiner, we refer to the Appeal Brief, filed July 26, 2019 (“Appeal. Br.”); the Reply Brief, filed Dec. 2, 2019 (“Reply Br.”); the Final Office Action, mailed Dec. 31, 2018 (“Final Act.”); and the Examiner’s Answer, mailed Oct. 1, 2019 (“Ans.”) for the respective details. Appeal 2020-001279 Application 14/755,401 4 ANALYSIS In rejecting the independent claims, the Examiner finds that Mekkat alone teaches the limitation, “determining an effective number of the multiple compute units to designate as having priority,” based on a method described in Mekkat for bypassing low level cache (LLC): [I]f the CPU application is cache sensitive, LowThr is selected for bypassing LLC for GPU memory accesses; otherwise, the choice of threshold depends on the characteristics of the GPU application. If GPU application is cache sensitive, HighThr is selected for bypassing LLC for GPU memory accesses. (Final Act. 3 (citing Mekkat, 229).) Appellant argues that this threshold selection algorithm does not determine an effective number of compute units to designate as having priority, but simply determines whether a particular application running on the particular compute unit will bypass cache. (Appeal Br. 7–8.) Appellant further argues: Mekkat is entirely concerned with dynamically determining whether the LLC should be bypassed by a particular GPU, and dynamically determining a threshold by which to make this determination. These determinations are made independently of other GPUs. Accordingly, Mekkat does not need to determine “an effective number” as in claim 1 — determining an effective number would be superfluous in Mekkat. In this sense, Mekkat in fact teaches away from “determining an effective number of the multiple compute units to designate as having priority” as in claim 1. (Appeal Br. 8.) In the Answer, the Examiner concludes that the broadest reasonable interpretation of “determining an effective number of the multiple compute units to designate as having priority” “only requires a determination that a Appeal 2020-001279 Application 14/755,401 5 sub-set (i.e. ‘a number of’) are effectively to be designated as having high priority at some point in the future.” (Ans. 4.) The Examiner bases this interpretation on the definition of “effective” as “producing a decided, decisive or desired effect.” (Id.) The Examiner further argues: Mekkat teaches a dynamic process wherein at a particular time there is an effective number of (i.e. a sub-set of) the multiple compute units to designate as having priority (i.e. enabled to bypass or not the LLC). What’s more, at any point in time, the effective number of compute units that are enabled to bypass the LLC is determined by the threshold. Thus, the threshold directly determines the effective number of compute units that are enabled to bypass the LLC. Hence, Mekkat clearly teaches the particular limitation. (Id.) Appellant replies, arguing that the plain meaning of “number” in the context of the claim requires a particular number of compute units to be determined and to be assigned priority, as supported by extensive disclosure in the specification describing dynamically determining the specific number of compute units to designate as having priority. (Reply 6–7.) Appellant further argues that the use of a threshold to determine whether an application can access cache is not a designation of a compute unit as having priority, as required by the claims. (Reply 7–8.) We agree with Appellant that the Examiner’s interpretation of “determining an effective number of the multiple compute units to designate as having priority” is unreasonably broad. The plain language of this claim limitation, as supported by the specification, requires determination of an Appeal 2020-001279 Application 14/755,401 6 actual number of compute units, and that those units be designated as having priority. Mekkat only discloses an algorithm which determines, for a particular application running on a particular compute unit, whether or not that applicator has access to cache. Therefore, we do not sustain the Examiner’s obviousness rejections of independent claims 1, 9, and 17 over the combination of Mekkat, Sim, and Hady, nor of claims 4–8, 12–16, 19, and 20, which depend from claim 1, 9 or 17. DECISION SUMMARY In summary: Claims Rejected 35 U.S.C. § Reference(s)/ Basis Affirmed Reversed 1, 4, 6, 7, 9, 12, 14, 15, 17, 19, 20 103 Mekkat, Sim, Hady 1, 4, 6, 7, 9, 12, 14, 15, 17, 19, 20 5, 13 103 Mekkat, Sim, Harriman 5, 13 8, 16 103 Mekkat, Sim, Brent 8, 16 Overall Outcome 1, 4–9, 12–17, 19, 20 REVERSED Copy with citationCopy as parenthetical citation