Ex Parte Al Sukhni et alDownload PDFPatent Trial and Appeal BoardSep 26, 201411262171 (P.T.A.B. Sep. 26, 2014) Copy Citation UNITED STATES PATENT AND TRADEMARK OFFICE ____________________ BEFORE THE PATENT TRIAL AND APPEAL BOARD ____________________ Ex parte HASSAN F. AL SUKHNI and JAMES C. HOLT ____________________ Appeal 2012–005131 Application 11/262,171 Technology Center 2100 ____________________ Before MICHAEL J. STRAUSS, CATHERINE SHIANG, and MINN CHUNG, Administrative Patent Judges. STRAUSS, Administrative Patent Judge. DECISION ON APPEAL Appeal 2012-005131 Application 11/262,171 2 STATEMENT OF THE CASE Appellants appeal under 35 U.S.C. § 134(a) from a rejection of claims 1, 3–10, and 22–27. Claims 2 and 11–21 are canceled. We have jurisdiction over the pending claims under 35 U.S.C. § 6(b). We affirm. THE INVENTION The claims are directed to decoupled precomputation prefetching of computer program instructions. Claim 1, reproduced below, is representative of the claimed subject matter: 1. A method comprising: generating, at a processing unit, a first prefetch graph based on a sequence of instructions of a program stream that are committed in an execution pipeline of the processing unit between a first iteration and a second iteration of a first load instruction of the program stream; and generating, at the processing unit, a second pre etch graph from the first prefetch graph based on a subset of the sequence of instructions that affect an address value associated with the first load instruction, wherein generating the second prefetch graph includes filtering out an instruction from the sequence of instructions based on a comparison of an instruction type of the instruction with an identified set of one or more instruction types. REFERENCES The prior art relied upon by the Examiner in rejecting the claims on appeal is: Wang (Wang ’083) 1 Wang (Wang ’101) Ahmad US 2002/0144083 A1 US 2002/0144101 A1 US 2004/0250164 A1 Oct. 3, 2002 Oct. 3, 2002 Dec. 9, 2004 1 Referenced as “Wang/Collins” and/or simply “Collins” in the Examiner’s Answer and Appellants’ Appeal Brief and Reply Brief. Appeal 2012-005131 Application 11/262,171 3 REJECTIONS2 The Examiner made the following rejections: Claims 22 and 26 stand rejected under 35 U.S.C. § 112 second paragraph as being indefinite. Ans. 4–6. Claims 1, 6–10, 22–24, 26, and 27 stand rejected under 35 U.S.C. § 103(a) as being unpatentable over Wang ’101 and Ahmad. Ans. 6–11. Claims 3–5 and 25 stand rejected under 35 U.S.C. § 103(a) as being unpatentable over Wang ’101, Ahmad, and Wang ’083. Ans. 11–13. ISSUES ON APPEAL Based on Appellants’ arguments in the Appeal Brief (App. Br. 4–12) and Reply Brief (Reply Br. 2–4), the issues presented on appeal are: 1. Whether the Examiner erred in concluding the recitation “executing the instruction loop independent of any triggering event” renders claims 22 and 26 indefinite under 35 U.S.C. § 112 second paragraph. 2. Whether the Examiner erred in finding Wang ’101’s multiple iterations of a criterion instruction teaches the disputed limitation of generating a prefetch graph based on a sequence of instructions 2 In connection with the rejections under 35 U.S.C. § 103(a) Appellants argue the rejection of independent claim 1 and dependent claim 4. Separate patentability is not argued for independent claims 24 or for dependent claims 3–10, 22, 23, and 25–27. “A statement which merely points out what a claim recites will not be considered an argument for separate patentability of the claim.” 37 C.F.R. § 41.37(c)(1)(vii)(last sentence). Further, merely restating with respect to a second claim an argument, previously presented with respect to a first claim, is not an argument for separate patentability of the two claims. Therefore, based on Appellants’ arguments, we decide the appeal of the rejection of claims 1, 3, 5–10, and 22–27 under 35 U.S.C. § 103(a) based on claim 1 alone and the appeal of claim 4 separately. See 37 C.F.R. § 41.37(c)(1)(vii). Appeal 2012-005131 Application 11/262,171 4 committed between iterations of a load instruction as required by claim 1. 3. Whether the Examiner erred in concluding the combination of Wang ’101’s generation of a prefetch graph by filtering instructions and Ahmad’s filtering instructions based on an instruction type teaches or suggests generating a second prefetch graph based on an instruction type as required by claim 1. 4. Whether the Examiner erred in combining the teachings of Wang ’101 and Ahmad in rejecting claim 1. 5. Whether the Examiner erred in concluding Wang ’101 and Ahmad when combined with Wang ’083’s general-purpose registers storing integer values evidence an integer load instruction type which is filtered from the instruction sequence thereby teaching or suggesting the disputed limitation of claim 4. ANALYSIS We have reviewed the Examiner’s rejections in light of Appellants’ arguments that the Examiner has erred. We disagree with Appellants’ conclusions. We adopt as our own (1) the findings and reasons set forth by the Examiner in the action from which this appeal is taken (Ans. 4–13) and (2) the reasons set forth by the Examiner in the Examiner’s Answer in response to Appellants’ Appeal Brief (Ans. 13–21) and concur with the conclusions reached by the Examiner. We highlight the following for emphasis. Appeal 2012-005131 Application 11/262,171 5 Rejection Under 35 U.S.C. § 112 Second Paragraph In connection with the rejections of claims 22 and 26 under 35 U.S.C. § 112 second paragraph, Appellants contend because “[o]ne skilled in the art understands that instructions included at a program stream have associated information, and that this information can trigger different events at an instruction pipeline,” the term “any triggering event” is definite. App. Br. 11. The Examiner finds “[a]ll part of the claimed invention are [at] least ‘associated with’ each other because they are all a part of the same invention [and t]herefore the scope of a ‘triggering event’ can seemingly include any source of action or control in the system.” Ans. 20. The Examiner concludes “[d]ue to such broad ‘triggering event’ scope, it is unclear how the instruction loop execution could possibly be ‘independent of any triggering event.’” Id. Absent sufficient evidence or argument in rebuttal to the Examiner’s rationale concluding the disputed language of claims 22 and 26 renders the claims indefinite, we find no error in the rejection under 35 U.S.C. § 112, second paragraph. Appellants’ argument in the Appeal Brief amounts little more than a general assertion of compliance with the requirements of 35 U.S.C. § 112, second paragraph (App. Br. 11) to which the Examiner responds with an explanation detailing specific inconsistencies found in the claims rendering the claims indefinite (Ans. 20–21). Because Appellants have not availed themselves of the opportunity to challenge in their Reply Brief the Examiner’s reasons for finding the disputed limitation indefinite as proffered in the Answer, the Examiner’s reasoning is unrebutted and, we find, otherwise reasonable. Therefore, we find no error in the Examiner’s conclusion of indefiniteness. Appeal 2012-005131 Application 11/262,171 6 In connection with the informalities found by the Examiner based on the language “associated with an instruction included at the instruction” recited in claim 26 (Ans. 6), other than acknowledge “a grammatical error” and assert the claim is not indefinite (App. Br. 12), Appellants make no substantive argument in response. Therefore, we find no error in the Examiner’s conclusion. For the reasons supra, we sustain the rejections of claims 22 and 26 under 35 U.S.C. § 112 second paragraph. Rejections Under 35 U.S.C. § 103(a) Claim 1 a. generating a prefetch graph based on a sequence of instructions committed between iterations of a load instruction Appellants contend Wang ’101’s directed acyclic graph (DAG) trace is determined by sliding an analysis window along an instruction sequence until a criterion instruction is found but is unrelated to and fails to teach a first iteration and a second iteration of a load instruction as required by claim 1. App. Br. 7. Appellants argue Wang ’101’s loop with multiple iterations of a criterion instruction fails to teach generating a prefetch graph based on a sequence of instructions committed between iterations of the instruction as required by claim 1. Id. The Examiner responds by finding Wang ’101’s criterion instruction can be a load instruction (Ans. 4) such that the DAG associated with the criterion instruction teaches a DAG related to a load instruction (id.). The Examiner further finds Wang ’101’s DAG is built and rebuilt during several iterations of the corresponding load and related instructions of a loop. Id. The Examiner concludes Wang ’101 teaches or Appeal 2012-005131 Application 11/262,171 7 suggests the disputed limitation of “generating, at a processing unit, a first prefetch graph based on a sequence of instructions . . . between a first iteration and a second iteration of a first load instruction of the program stream.” Ans. 7, 14. We agree with the Examiner. Appellants’ argument addressing Wang ’101’s sliding analysis window is unpersuasive because it does not address the Examiner’s findings. Furthermore, because Wang ’101’s criterion instruction is associated with the DAG, under a broad but reasonable interpretation of the recited “based on” relationship between the prefetch graph and load instruction of claim 1, and absent sufficient evidence or argument to the contrary, we agree with the Examiner Wang ’101’s DAG associated with the criterion instruction teaches or suggests a prefetch graph based on a sequence of instructions including a load instruction. Because Wang ’101’s sequence of instructions forms a loop that is iteratively executed with the DAG being built and rebuilt, we also agree the reference teaches or suggests the entirety of the disputed limitation. b. generating a second prefetch graph based on an instruction type Appellants contend Ahmad is limited to filtering information sent to an external debug system via a trace port and fails to teach filtering an instruction to generate a prefetch graph as required by claim 1. App. Br. 8. Appellants argue the Examiner’s finding that “filtering and fine-tuning program code is extremely common in the art” (App. Br. 9; Ans. 12) is not a proper basis for a conclusion of obviousness and “the fact that some types of filtering or fine-tuning may be known or may be desirable does not imply that the recited features are also known in the art” (App. Br. 9). The Appeal 2012-005131 Application 11/262,171 8 Examiner responds by explaining the disputed limitation, while not disclosed by the references individually, is taught or suggested by the combination of Wang ’101 and Ahmad. Ans. 15. In particular, the Examiner finds “Wang teaches generating a second prefetch graph by filtering instructions, while Ahmad teaches filtering instructions based on an instruction type.” Ans. 15–16. The Examiner concludes it would have been obvious to modify Wang ’101 to filter instructions based on instruction type as taught by Ahmad to provide an additional tool to optimize the instructions in a DAG, reduce storage requirements, and improve speed during DAG rebuilding. Ans. 16–17. We are unpersuaded of error by Appellants’ arguments. Appellants are arguing the teachings of Ahmad separately without considering the combined teachings of Wang ’101 and Ahmad. “Non-obviousness cannot be established by attacking references individually where the rejection is based upon the teachings of a combination of references.” In re Merck & Co., 800 F.2d 1091, 1097 (Fed. Cir. 1986) (citing In re Keller, 642 F.2d 413, 425 (CCPA 1981)). Appellants have not explained why it would not be obvious to use the filtering of Ahmad in the system of Wang ’101 or presented evidence the combination would not result in teaching or suggesting the disputed limitation. In contrast, we find the Examiner’s findings and conclusion of obviousness to be supported by the evidence of record. Therefore, in the absence of sufficient evidence, Appellants’ contention is not persuasive of Examiner error. Appeal 2012-005131 Application 11/262,171 9 c. One skilled in the art would not combine the references as proposed Appellants contend “[n]either Wang [’101] nor Ahmad provide any motivation for combining their disparate teachings [nor is there any] teaching in either reference as to how the two systems would interact or combine to reduce storage space or improve speed, as asserted by the Office.” App. Br. 9. Appellants argue “[c]ombining the debug system of Ahmad with the DAG trace system of Wang would not reduce the storage space or improve the speed of the DAG trace system in any manner.” Id. The Examiner responds “the test for obviousness is not whether the features of a secondary reference may be bodily incorporated into the structure of the primary reference. Rather, as discussed above, the test is what the combined teachings of the references would have suggested to those of ordinary skill in the art.” Ans. 18. We agree with the Examiner. The Examiner explains use of Ahmad’s filtering of instruction based on instruction type in Wang ’101’s system would reduce storage requirement and improve speed. Although Appellants contest the Examiner’s findings, insufficient evidence is provided to persuade us the Examiner has erred.3 In particular, Appellants fail to provide sufficient evidence or argument that the combination yields anything other than predictable results. See KSR Int’l Co. v. Teleflex, Inc., 550 U.S. 398, 416 (2007) (“The combination of familiar elements according to known methods is likely to be obvious when it does no more than yield predictable results.”). We are also not persuaded that combining the respective familiar 3 Attorney arguments and conclusory statements that are unsupported by factual evidence are entitled to little probative value. See In re Geisler, 116 F.3d 1465, 1470 (Fed. Cir. 1997); see also In re De Blauwe, 736 F.2d 699, 705 (Fed. Cir. 1984.) Appeal 2012-005131 Application 11/262,171 10 elements of the cited references in the manner proffered by the Examiner was “uniquely challenging or difficult for one of ordinary skill in the art.” Leapfrog Enters., Inc. v. Fisher-Price, Inc., 485 F.3d 1157, 1162 (Fed. Cir. 2007) (citing KSR, 550 U.S. at 418). Therefore, we find the Examiner’s proffered combination of familiar prior art elements according to their established functions would have conveyed a reasonable expectation of success to a person of ordinary skill at the time of the invention. We further find the Examiner has articulated reasoning with rational underpinnings sufficient to justify the legal conclusion of obviousness. Therefore, we find unpersuasive of error Appellants’ contention that the combination of Wang ’101 and Ahmad is improper. For the reasons supra, Appellants’ arguments are unpersuasive of Examiner error. Accordingly, we sustain the rejections of independent claim 1 and, for the same reasons, independent claim 24 under 35 U.S.C. § 103(a) over Wang ’101 and Ahmad together with the rejection of dependent claims 6–10, 22, 23, 26, and 27 not separately argued. We further sustain the rejection of dependent claims 3, 5, and 25 under 35 U.S.C. § 103(a) over Wang ’101, Ahmad, and Wang ’083, these dependent claims also not separately argued. Claim 4 In connection with claim 4, Appellants contend even if Wang ’083’s instructions using a general purpose register were to imply integer operations, “there is no disclosure in any of the cited references of filtering based on an integer load instruction type.” App. Br. 11. The Examiner responds by finding Wang ’083 (i.e., “Wang/Collins” or “Collins”) teaches filtering both add and load instruction types, both of which can be an integer Appeal 2012-005131 Application 11/262,171 11 type. Ans. 19. The Examiner concludes claim 4 is unpatentable over the combination of Wang ’101, Ahmad, and Wang ’083. We are not persuaded of error. Although Appellants reply by asserting Figure 6 of Wang ’083 “discloses only that an integer load instruction can be used in a sequence of instructions” but not “that filtering was performed based on the integer load instruction type in any manner” (Reply Br. 4), Appellants do not address the Examiner’s specific findings. In particular, the Examiner finds Figure 6 of Wang ’083 depicts “[an] exemplified p-slice subset comprises load and add instructions that are filtered from the instruction sequence; the other load and add instructions of sequence 602 can be seen to have been filtered out when creating the p-slice.” Ans. 19. The Examiner further finds the Figure “explicitly shows filtering an integer load (program 602 line 404991: ‘ld4 r21 =[r15]’; this load instruction is of the integer type because it loads a value into integer register ‘r21’; that is, ‘r21’ is shorthand for general-purpose integer register 21; this instruction is filtered based on its type – a delinquent integer load instruction ).” Id. Absent sufficient evidence or argument in rebuttal, we find the Examiner’s rationale to be reasonable, as is the conclusion of obviousness. Appellants’ contention is further unpersuasive of error because the Examiner relies on the combination of Wang ’101 and Ahmad, not Wang ’083, for teaching or suggesting filtering based on a load instruction type (Ans. 7, 11) as required by claim 3. Wang ’083 is included for the specific teaching of an integer load instruction type. Ans. 12. Therefore, Appellants’ argument asserting Wang ’083 fails to teach filtering based on an integer load instruction type (App. Br. 11) is not persuasive of error because Appellants are arguing the teachings of Wang ’083 separately without Appeal 2012-005131 Application 11/262,171 12 considering the combined teachings of Wang ’101, Ahmad, and Wang ’083. “Non-obviousness cannot be established by attacking references individually where the rejection is based upon the teachings of a combination of references.” Merck, 800 F.2d at 1097 (citing Keller, 642 F.2d at 425). For the reasons supra, Appellants’ contentions are not persuasive of Examiner error. Therefore, we sustain the rejection of claim 4 under 35 U.S.C. § 103(a) over Wang ’101, Ahmad, and Wang ’083. CONCLUSIONS 1. The Examiner did not err in concluding the recitation “executing the instruction loop independent of any triggering event” renders claims 22 and 26 indefinite under 35 U.S.C. § 112, second paragraph. 2. The Examiner did not err in finding Wang ’101’s multiple iterations of a criterion instruction teaches the disputed limitation of generating a prefetch graph based on a sequence of instructions committed between iterations of a load instruction as required by claim 1. 3. The Examiner did not err in concluding the combination of Wang ’101’s generation of a prefetch graph by filtering instructions and Ahmad’s filtering instructions based on an instruction type teaches or suggests generating a second prefetch graph based on an instruction type as required by claim 1. 4. The Examiner did not err in combining the teachings of Wang ’101 and Ahmad in rejecting claim 1. 5. The Examiner did not err in concluding Wang ’101 and Ahmad when combined with Wang ’083’s general-purpose registers storing integer values evidence an integer load instruction type is filtered from the Appeal 2012-005131 Application 11/262,171 13 instruction sequence thereby teaching or suggesting the disputed limitation of claim 4. DECISION The Examiner’s decision to reject claims 1, 3–10, and 22–27 is affirmed. No time period for taking any subsequent action in connection with this appeal may be extended under 37 C.F.R. § 1.136(a)(1)(iv). AFFIRMED msc Copy with citationCopy as parenthetical citation