Ex Parte Doing et alDownload PDFPatent Trial and Appeal BoardApr 23, 201412176386 (P.T.A.B. Apr. 23, 2014) 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. 12/176,386 07/20/2008 Richard William Doing RPS920040209US3 2523 10584 7590 04/24/2014 Winstead, P.C. P.O.Box 131851 Dallas, TX 75313 EXAMINER LI, AIMEE J ART UNIT PAPER NUMBER 2183 MAIL DATE DELIVERY MODE 04/24/2014 PAPER 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. PTOL-90A (Rev. 04/07) UNITED STATES PATENT AND TRADEMARK OFFICE ____________ BEFORE THE PATENT TRIAL AND APPEAL BOARD ____________ Ex parte RICHARD WILLIAM DOING, BRETT OLSSON, and KENICHI TSUCHIYA ____________ Appeal 2011-010612 Application 12/176,386 Technology Center 2100 ____________ Before ROBERT E. NAPPI, BRUCE R. WINSOR, and LINZY T. McCARTNEY, Administrative Patent Judges. McCARTNEY, Administrative Patent Judge. DECISION ON APPEAL Appellants appeal under 35 U.S.C. § 134(a) the Examiner’s final rejection of claims 1, 2, 4, 5, 9-11, and 17. Claims 6 and 7 are canceled, and claims 3, 8, 12-16, and 18 are allowed. We have jurisdiction under 35 U.S.C. § 6(b). We AFFIRM. Appeal 2011-010612 Application 12/176,386 2 STATEMENT OF THE CASE Appellants’ invention relates to “reducing the fetch time of target instructions of a predicted taken branch instruction.” Spec. 1. Claim 1 illustrates the claimed subject matter: 1. A method for reducing the normal fetch time of a predicted taken branch instruction comprising the steps of: accessing an instruction cache to fetch an instruction; indexing into an entry in a buffer using bits from said instruction fetched from said instruction cache, wherein said buffer comprises a plurality of entries, wherein each of said plurality of entries comprises an address of a branch instruction, a plurality of instructions beginning at a target address of said branch instruction, prediction information for any of said plurality of instructions that are branch instructions and an address of a next fetch group; comparing an address of said instruction fetched from said instruction cache with said address of said branch instruction in said indexed entry of said buffer; and selecting said plurality of instructions beginning at said target address of said branch instruction in said indexed entry of said buffer if said address of said instruction fetched from said instruction cache matches with said address of said branch instruction in said indexed entry of said buffer. REJECTION Claims 1, 2, 4, 5, 9-11, and 17 stand rejected under 35 U.S.C. § 103(a) as unpatentable over McDonald (US 2004/0143709 Al; July 22, 2004) in view of Grohoski (US 5,794,027; August 11, 1998). Appeal 2011-010612 Application 12/176,386 3 ISSUES Appellants’ Appeal Brief presents the following issues1: 1. Does the cited art teach or suggest the “entries,” “comparing,” and “selecting” limitations recited in claim 1? 2. Does the cited art teach or suggest the “selecting” limitation recited in claim 2? 3. Does the cited art teach or suggest the “selecting” limitation recited in claim 4? 4. Does the cited art teach or suggest the “determining” limitation recited in claim 5? 5. Does the cited art teach or suggest “a second selection mechanism” as recited in claim 11? 6. Does the cited art teach or suggest the “buffer queue” recited in claim 17? 7. Did the Examiner provide sufficient supporting rationale for the obviousness rejections of claims 1, 2, 4, 5, 11, and 17? 1 Appellants’ Reply Brief includes arguments not presented in Appellants’ Appeal Brief. Appellants have waived these arguments by failing to raise them in their Appeal Brief. See Ex parte Borden, 93 USPQ2d 1473, 1474 (informative) (“[T]he reply brief [is not] an opportunity to make arguments that could have been made in the principal brief on appeal to rebut the Examiner’s rejections . . . .”). See also Becton Dickinson & Co. v. C.R. Bard, Inc., 922 F.2d 792, 800 (Fed. Cir. 1990) (“[A]n issue not raised by an appellant in its opening brief . . . is waived.”). Accordingly, we consider only arguments Appellants properly raised in their Appeal Brief. Appeal 2011-010612 Application 12/176,386 4 ANALYSIS Claim 1 “Entries” Claim 1 recites a buffer including a “plurality of entries,” each entry containing the following four components: (1) “an address of a branch instruction,” (2) “a plurality of instructions beginning at a target address of said branch instruction,” (3) “prediction information for any of said plurality of instructions that are branch instructions,” and (4) “an address of a next fetch group.” The Examiner found McDonald taught the recited “entries.” As shown in Figure 3 below, McDonald discloses a Branch Target Address Cache (referred to as a “BTAC”) that includes target address, tag, and counter arrays: Appeal 2011-010612 Application 12/176,386 5 Figure 3: BTAC Diagram Figure 3 shows an address 182 selecting lines in the target address, tag, and counter arrays. The Examiner found the selected lines constituted a single “entry,” and determined these lines, when combined with a branch buffer storing a “plurality of instructions beginning at a target address” taught by Grohoski, contained the four components recited in claim 1. (Ans. at 4-6.) In response, Appellants argue the Examiner’s interpretation of the recited “entries” is inconsistent with McDonald’s use of the word. Appeal 2011-010612 Application 12/176,386 6 Appellants note McDonald refers to individual storage elements as “entries” (for example, target address array entry 312 in Figure 3 above), not multiple storage elements as interpreted by the Examiner. (See Reply Br. at 1-4.) Under this narrower interpretation, Appellants contend none of the disclosed storage elements includes all four of the recited components. (Id.) Moreover, Appellants seemingly argue even under the Examiner’s interpretation of “entries,” McDonald neither teaches nor suggests an entry including the fourth recited component, “an address of a next fetch group.” (Id. at 4.) We agree with the Examiner. During prosecution, we give claim terms their broadest reasonable interpretation in light of the applicant’s specification. In re Morris, 127 F.3d 1048, 1054 (Fed. Cir. 1997). Here, Appellants’ specification discloses an illustrative “entry” that includes multiple entries, each entry generally containing multiple bits. See, e.g., Spec. 12 (“BTB entry 400 may include entries 401A-L. Entry 401A may store bits 0-22 and 27-29 of the branch instruction address. Entry 401B may store an address of the target address (bits 0-29) . . . .”). Far from precluding the Examiner’s interpretation of “entry,” the illustrative entry indicates an “entry” includes more than one storage element. Even absent this disclosure, Appellants have pointed to nothing in either the claims or their specification that limits “entry” to a single storage element. Although prior art such as McDonald can shed light on the meaning of claim terms, we note the cited art also suggests the broadest reasonable interpretation of “entry” includes multiple storage elements. Grohoski, another prior art reference cited by the Examiner, discloses a Branch Anticipate Buffer (known as a “BAB”) that includes a table of addresses and Appeal 2011-010612 Application 12/176,386 7 instructions beginning at each address. (Grohoski, Figure 1, col. 3, ll. 56- 67.) Grohoski refers to each line of the table as an “entry.” (See id.) “[T]he fact that appellants can point to definitions or usages that conform to their interpretation does not make the PTO’s definition unreasonable when the PTO can point to other sources that support its interpretation.” Morris, 127 F.3d at 1056. As for the “address of a next fetch group,” Appellants apparently contend the cited art does not teach an address of a group of instructions. (Reply Br. at 4.) Appellants also contend the Examiner’s rejection relies on a “BTAC write queue,” which Appellants assert the Examiner erroneously found was part of the BTAC. (App. Br. at 9.) We disagree. As an initial matter, the Examiner’s Answer made clear the Examiner’s rejection did not turn on the BTAC write queue being part of the BTAC. Rather, as explained in the Answer, the Examiner found McDonald taught fetching multiple instructions at a target address. (Ans. at 4-5, 12); McDonald ¶ 0007 (“[T]he processor must . . . begin fetching instructions at the target address . . . .” (emphasis added)). The Examiner also found McDonald disclosed target addresses were stored in the BTAC and served as “next current fetch addresses.” (Ans. at 4-5, 12.) Based on these findings, the Examiner concluded the disclosed target addresses were the next fetch addresses of a group of instructions. (Id. at 12.) Appellants have not convinced us these findings are erroneous. “Comparing” The Examiner determined the combination of McDonald and Grohoski taught “comparing an address of said instruction fetched from said instruction cache with said address of said branch instruction in said indexed Appeal 2011-010612 Application 12/176,386 8 entry of said buffer.” The Examiner found McDonald disclosed comparing an address with “tags” that each included part of a branch instruction address. (Ans. at 5.) The Examiner acknowledged McDonald did not teach the address came from the recited “instruction cache” but found Grohoski taught fetching instructions from an instruction cache. (Id. at 5-6.) In their Appeal Brief, Appellants simply argue the Examiner has not pointed to any language in either McDonald or Grohoski that suggests fetching an address from an instruction cache. (App. Br. at 10-11.) Appellants argue for the first time in their Reply Brief that the cited art does not teach “comparing an address of an instruction fetched with the address of the branch instruction” or “in the indexed entry of the buffer.” (Reply Br. at 6-7.) We agree with the Examiner. First, Appellants essentially argue neither McDonald nor Grohoski individually teaches or suggests fetching an address from an instruction cache. But the Examiner determined the combination of McDonald and Grohoski suggested this limitation, and Appellants’ cursory arguments have not convinced us the Examiner’s determination was erroneous. See In re Keller, 642 F.2d 413, 426 (CCPA 1981) (“[O]ne cannot show non-obviousness by attacking references individually where, as here, the rejections are based on combinations of references.”). Second, as alluded to earlier, Appellants have waived their remaining arguments concerning this limitation because Appellants failed to raise the arguments in the Appeal Brief. See Borden, 93 USPQ2d at 1474; Becton Dickinson & Co., 922 F.2d at 800. Regardless, we note that these arguments are meritless. As explained by the Examiner, McDonald discloses comparing an address with tags that contain branch instruction Appeal 2011-010612 Application 12/176,386 9 addresses. (See, e.g., Ans. at 5.) We agree this disclosure, combined with the teachings of Grohoski, suggests this limitation. “Selecting” The Examiner found the combination of McDonald and Grohoski suggested the “selecting” step recited in claim 1. Appellants argue the cited combination does not teach or suggest selecting anything “if said address of said instruction fetched from said instruction cache matches with said address of said branch instruction in said indexed entry of said buffer” as recited in claim 1. (App. Br. at 11-14.) The use of the conditional word “if” makes the “selecting” limitation optional; therefore, this limitation is not entitled to patentable weight and the Examiner need not find prior art that teaches this limitation. Cf. Ex Parte Katz, 2010-006083, 2011 WL 514314, at *4-5 (BPAI 2011) (non- precedential) (concluding that, under the broadest reasonable interpretation of the claim at issue, an examiner was not required to find an “if” limitation in the prior art). Accordingly, we affirm the Examiner’s rejection of claim 1. We note, however, that we agree with the Examiner the combination of McDonald and Grohoski suggests the limitation for the reasons set forth in the Examiner’s Final Rejection and Answer. Claims 2 and 4 Claims 2 and 4 recite selecting an instruction or information, respectively, “if” a specified condition occurs. Appellants contend the combination of McDonald and Grohoski neither teaches nor suggests these limitations. (App. Br. at 14-21.) As with the “selecting” limitation in claim 1, these limitations are optional and therefore the Examiner need not find these limitations in the prior art. Cf. Katz, 2011 WL 514314, at *4-5. Thus, Appeal 2011-010612 Application 12/176,386 10 even if Appellants were correct the cited art failed to teach or suggest these limitations, we would not be persuaded the Examiner’s rejection of these claims was erroneous. Claim 5 Claim 5 recites “determining if any of said plurality of instructions selected is a branch instruction.” The Examiner found the combination of McDonald and Grohoski disclosed the recited subject matter, in particular Figure 3D of Grohoski. (Ans. at 7.) Appellants argue the Examiner overlooked the recited “said plurality of instructions” are “select[ed] said plurality of instructions beginning at said target address of said branch instruction in said indexed entry of said buffer,” as claim 5 depends from—and therefore incorporates the limitations of—claim 1. Appellants contend Grohoski does not teach or suggest this limitation. (App. Br. at 19-21.) We agree with the Examiner. Because claim 5’s “determining” step depends upon claim 1’s “selecting” step occurring, Appellants correctly argue the “selecting” step is entitled to patentable weight when considering claim 5. But as noted above, the Examiner found the combination of McDonald and Grohoski suggested the “selecting” limitation recited in claim 1. The Examiner also found step 101 of Grohoski’s Figure 3D discloses a decision branch that determines whether an instruction is a branch instruction. (See Grohoski, Figure 3D.) The description accompanying Figure 3D teaches comparing a branch target address to the BAB, and if the address is in the BAB, copying the contents of the BAB into a sequential buffer. (Id. at col. 6, ll. 52-55.) If the system has not resolved the branch instruction, the system determines whether the next instruction in Appeal 2011-010612 Application 12/176,386 11 the sequential buffer is a branch instruction. (Id. at col. 7, ll. 2-7.) We see no error in the Examiner’s finding that these teachings suggest the disputed limitation. Claim 11 Claim 11 recites “a second selection mechanism coupled to said first selection mechanism, wherein said second selection mechanism is configured to select an address of one of said plurality of instructions selected to be loaded into an instruction queue.” The Examiner found Grohoski disclosed an instruction fetch mechanism that suggested this limitation, in particular, the instruction cache, BAB, and target buffer of the mechanism. (Ans. at 8.) Figure 2 of Grohoski illustrates an embodiment of the mechanism: Figure 2: Block Diagram of an Instruction Fetch Mechanism Appellants argue the cited art does not teach a “second selection mechanism coupled to the first selection mechanism” or “where the second Appeal 2011-010612 Application 12/176,386 12 selection mechanism is configured to select an address of one of the plurality of instructions selected to be loaded into an instruction queue.” (App. Br. at 21-22.) Moreover, Appellants contend the Examiner has not clarified which of the target buffer and sequential buffer shown in Figure 2 serves as the claimed first and second “selection mechanisms.” (Id. at 22.) Appellants misapprehend the Examiner’s rejection. As explained in the Examiner’s Final Rejection, the Examiner found the target buffer was the recited “instruction queue” and the instruction cache and BAB were the first and second “selection mechanisms.” (Ans. at 8, 18.) As also noted by the Examiner, the BAB can place selected instructions in either the target or the sequential buffers. (Id. at 18.) We agree with the Examiner that this mechanism teaches the claimed subject matter. Claim 17 Claim 17 recites “a buffer queue coupled to said buffer, wherein said buffer queue is configured to store instructions selected from said instruction cache, wherein said buffer queue is further configured to store prediction information selected from a branch history table.” The Examiner found the proposed McDonald-Grohoski combination suggested this limitation, specifically the sequential buffer, BAB, instruction cache, and target buffer depicted in Figure 2 of Grohoski. (Ans. at 9.) Appellants contend Grohoski does not disclose the sequential buffer shown in Figure 2 is configured to store prediction information selected from the target buffer. (App. Br. at 24.) Appellants again argue against the cited references individually, but the Examiner determined the combination of McDonald and Grohoski suggested the limitations recited in claim 17. The Examiner found that Appeal 2011-010612 Application 12/176,386 13 McDonald disclosed that target address array entries included prediction information, (Ans. at 4, 7-8), and concluded the combined McDonald- Grohoski system would store those entries and their associated prediction information in the sequential buffer, (id.). Appellants have not persuaded us this finding is erroneous. Motivation to Combine The Examiner found that one of ordinary skill in the art would have been motivated to combine McDonald and Grohoski in the proposed manner “for the purpose of improving the overall system performance . . . by minimizing processor idle time during instruction cache miss or branch misprediction.” (Ans. at 6.) The Examiner also found Appellants’ alleged invention was simply a combination of familiar components using known methods that yielded predictable results, and the combination was obvious to try. (Id. at 19-23.) Appellants attack this finding on a number of grounds: Appellants contend the Examiner’s motivation stems from a misstatement of Grohoski. (See, e.g., App. Br. at 27.) Appellants argue the Examiner failed to explain why Grohoski, when accurately read, would motivate one of skill in the art to make the proposed changes. (Id.) Moreover, Appellants contend the Examiner failed to make a number of factual findings Appellants believe are necessary to support the obviousness rejection. (See, e.g., id. at 30, 33-34.) Appellants also argue the proposed modifications would change McDonald’s principle of operation. (Id. at 34-35.) Appellants’ arguments are unpersuasive. First, the Examiner did not misstate Grohoski. Rather, as indicated by the “see” signal used by the Examiner, the Examiner inferred from Grohoski’s “objects of the invention” Appeal 2011-010612 Application 12/176,386 14 statements that Grohoski’s system also aimed to minimize idle time during instruction cache reloading when a cache miss or misprediction occurs. (See, e.g., Ans. at 6.) Other portions of Grohoski support this inference. (See, e.g., Grohoski col. 6, ll. 55-61 (“This provides a temporary storage for the sequential buffer’s contents in the event that the branch prediction logic proves to be wrong . . . . This again shows how the BAB improves performance by immediately providing an instruction stream to the sequential buffer without having to access the I-cache.” (emphasis added))); (see also id. at col. 2, l. 56 – col. 3, l. 13.) Other than pointing out the Examiner did not quote Grohoski, Appellants have not seriously challenged this finding. Second, the Examiner provided “articulated reasoning with some rational underpinning” to support the obviousness rejection. The Examiner found McDonald and Grohoski disclosed similar inventions, (Ans. at 6), and, as just discussed, the Examiner inferred that one goal of Grohoski’s system was to minimize processor idle time in certain situations, (id. at 6). Based on these findings, the Examiner found it would have been obvious to make the proposed combination to improve overall system performance and minimize processor idle time in specified contexts. (See id.) We agree with these findings. To conclude otherwise, we would have to ignore common sense and deprive one of skill in the art the creative inferences of an ordinary person. That we cannot do. See KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398, 421 (2007) (“A person of ordinary skill is also a person of ordinary creativity, not an automaton.”). The Examiner bolstered this determination in the Answer by citing the Supreme Court’s statement that “[t]he combination of familiar elements Appeal 2011-010612 Application 12/176,386 15 according to know methods is likely to be obvious when it does no more than yield predictable results.” (Ans. at 20 (citing KSR, 550 U.S. 398).) The Examiner also found it would have been obvious to try certain modifications. (Id. at 21.) We need not consider whether the Examiner made the factual findings necessary to support this additional reasoning because the Examiner’s main rationale for combining the references— improving overall system performance and minimizing processor idle time—is sufficient to support the rejection. Finally, Appellants have not persuaded us the proposed modifications would change McDonald’s principle of operation. Appellants argue that if, as proposed by the Examiner, the address 182 shown in Figure 3 above originated from an instruction cache instead of a multiplexer, McDonald’s system would no longer be able to avoid deadlock conditions. But Appellants provide no supporting evidence or rationale for this assertion, and it is unclear why changing the location the address originates from and not the address itself would change how McDonald’s system functions. Regardless, “it is not necessary that the inventions of the references be physically combinable to render obvious the invention under review.” In re Sneed, 710 F.2d 1544, 1550 (Fed. Cir. 1983). DECISION We have considered Appellants’ remaining arguments and found them either waived, unpersuasive, or both. As the other rejected claims stand or fall with the claims discussed above, we affirm the Examiner’s final rejection of claims 1, 2, 4, 5, 9-11, and 17 under 35 U.S.C. § 103(a). Appeal 2011-010612 Application 12/176,386 16 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) (2011). See 37 C.F.R. § 1.136(a)(1)(iv). AFFIRMED msc Copy with citationCopy as parenthetical citation