Ex Parte Tabony et al

Patent Trial and Appeal Board

Mar 27, 2017

13449754 (P.T.A.B. Mar. 27, 2017)

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.
13/449,754 04/18/2012 Charles Joseph Tabony 110446 3707
23696 7590 03/29/2017
OTTAT mMM TNmRPORATFD
EXAMINER
5775 MOREHOUSE DR.
SAN DIEGO, CA 92121
GIROUX, GEORGE
ART UNIT PAPER NUMBER
2182
NOTIFICATION DATE DELIVERY MODE
03/29/2017 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):
us-docketing@qualcomm.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Ex parte CHARLES JOSEPH TABONY, LUCIAN CODRESCU,
and SURESH K. VENKUMAHANTI
Appeal 2017-000747
Application 13/449,7541
Technology Center 2100
Before ERIC S. FRAHM, JOHN D. HAMANN, and
MATTHEW J. McNEILL, Administrative Patent Judges.
HAMANN, Administrative Patent Judge.
DECISION ON APPEAL
Appellants file this appeal under 35 U.S.C. § 134(a) from the
Examiner’s Final Rejection of claims 1—36. We have jurisdiction under
35 U.S.C. § 6(b). We affirm.
THE CLAIMED INVENTION
Appellants’ claimed invention relates to branch prediction
mechanisms in computer processors, including predicting the outcome of
instructions and encoding the predictions in the instructions. Spec. 1—2.
1 According to Appellants, the real party in interest is QUALCOMM
Incorporated. App. Br. 3.
Appeal 2017-000747
Application 13/449,754
Claim 1 is illustrative of the subject matter of the appeal and is reproduced
below.
1. A method of predicting evaluation of a producer
instruction comprising:
encoding a prediction field in the producer instruction;
and
predicting evaluation of the producer instruction, in a
processor, using the prediction field.
REJECTIONS ON APPEAL
(1) The Examiner rejected claims 1—7, 10-16, 18—21, 24—31, and
34—36 under 35 U.S.C. § 103(a) as being unpatentable over the combination
of Grochowski et al. (US 2003/0126414 Al; published July 3, 2003)
(hereinafter “Grochowski”) and Gschwind (US 2008/0005542 Al; published
Jan. 3, 2008).
(2) The Examiner rejected claims 8, 9, 22, 23, 32, and 33 under
35 U.S.C. § 103(a) as being unpatentable over the combination of
Grochowski, Gschwind, and well-known practices in the art.
(3) The Examiner rejected claim 17 under 35 U.S.C. § 103(a) as
being unpatentable over the combination of Grochowski, Gschwind, and
Hoyt et al. (US 5,584,001; issued Dec. 10, 1996) (hereinafter “Hoyt”).
ANALYSIS
We have reviewed the Examiner’s rejections in light of Appellants’
contentions that the Examiner erred. In reaching our decision, we consider
all evidence presented and all arguments made by Appellants.
We disagree with Appellants’ arguments and we incorporate herein
and adopt as our own the findings, conclusions, and reasons set forth by the
2
Appeal 2017-000747
Application 13/449,754
Examiner in the (1) July 17, 2015 Final Office Action (“Final Act.” 2—15),
(2) November 10, 2015 Advisory Action (“Adv. Act.” 2—3), and (3) August
26, 2016 Examiner’s Answer (“Ans.” 2—17). We highlight and address,
however, specific findings and arguments below for emphasis.
(1) Encoding a prediction field in the producer instruction
Appellants argue the combination of Grochowski and Gschwind, and
Grochowski in particular, fails to teach or suggest “predicting evaluation of
the producer instruction, in a processor, using the prediction field,” as
recited in claim 1, and similarly recited in claims 12, 26, and 34. Reply Br.
3^4. According to Appellants, Grochowski does not teach predicting a
producer instruction (i.e., the compare instruction which is the producer
instruction of predicate registers p3 and p4), but instead teaches predicting
the value of predicate registers (e.g., p3, p4). See Reply Br. 3 (citing
Grochowski || 12—13, Table 1). Appellants contend Grochowski, thus,
instead teaches the opposite of claim 1.
The Examiner finds, and we agree, the combination of Grochowski
and Gschwind teaches or suggests the disputed limitation. Ans. 12—14; Adv.
Act. 2—3. As to Grochowski, the Examiner finds, and we agree, it teaches or
suggests “predicting the outcome of a producer compare instruction using
prediction information.” See Ans. 12—13 (citing Grochowski 113); Adv.
Act. 2—3; see also Ans. 14 (citing Grochowski || 12—13). For example,
Grochowski teaches “the past history of the compare instruction 011 is used
to predict the value of predicate registers p3 and p4 before the instruction
011 is executed,” where compare “instruction 011 is a producer of predicate
registers, p3 and p4.” Grochowski || 12—13; see also Grochowski 113
(teaching constructing a history or pattern table of past evaluations for use in
3
Appeal 2017-000747
Application 13/449,754
predicting, inter alia, values of predicate registers). We find Appellants’
purported dichotomy (i.e., predicting a producer instruction versus
predicting the value of a predicate register) unpersuasive, including because
the claim recites “predicting evaluation of the producer instruction” and
such evaluation is what is used to set the value of the predicate register. See
Spec. 16 (“Processors which adopt the use of predicate registers may
include instructions to generate the values for the predicate registers,
referred to herein as ‘producer instructions.’ . . . [Producer instructions
. . . may be used to write or set the predicate registers.”). Predicting
evaluation of a producer instruction and predicting the value of the
associated predicate register are one and the same.
(2) Encoding a prediction field in the producer instruction
Appellants argue the combination of Grochowski and Gschwind, and
Gschwind in particular, fails to teach or suggest “encoding a prediction field
in the producer instruction,” as recited in claim 1, and similarly recited in
claims 12, 26, and 34. App. Br. 5—7; Reply Br. 3^4. More specifically,
Appellants argue Gschwind teaches, at best, a bimodal predictor within a
consumer instruction — not a producer instruction as claimed. App. Br. 5—6
(citing Gschwind 17); see also Spec. 6 (“The one or more instructions,
such as conditional branch instructions, which employ the predicate registers
are referred to herein as ‘consumer instructions.’ The consumer instructions
are said to be predicated on the producer instructions.”).
The Examiner finds, and we agree, the combination of Grochowski
and Gschwind teaches or suggests the disputed limitation. Ans. 12—14; Adv.
Act. 2. As discussed in more detail above, the Examiner finds, and we
agree, Grochowski teaches or suggests, inter alia, predicting the outcome of
4
Appeal 2017-000747
Application 13/449,754
producer instructions using prediction information. Ans. 12—14 (citing
Grochowski || 12—13); Adv. Act. 2—3. The Examiner also finds, and we
agree, Grochowski teaches a history or table of past evaluations for use in
making predictions concerning, inter alia, producer instructions. See, e.g.,
Adv. Act. 3 (citing Grochowski 113). As to Gschwind, the Examiner finds,
and we agree, it teaches or suggests that a bimodal branch prediction (i.e., a
prediction field) or hint bits can be stored in the instruction it is being used
to predict. Ans. 14 (citing Gschwind 117).
We agree with the Examiner that the combined teachings of
Grochowski and Gschwind teach, or at least suggest, encoding Gschwind’s
prediction field within the producer instruction being predicted, as taught by
Grochowski. Id.; Adv. Act. 2—3; KSR Int’l Co. v. Teleflex Inc., 550 U.S.
398, 418 (2007) (“[T]he [obviousness] analysis need not seek out precise
teachings directed to the specific subject matter of the challenged claim, for
a court can take account of the inferences and creative steps that a person of
ordinary skill in the art would employ.”); see also In re Preda, 401 F.2d 825,
826 (CCPA 1968) (“[I]t is proper to take into account not only specific
teachings of the references but also the inferences which one skilled in the
art would reasonably be expected to draw therefrom.”). We find Appellants’
arguments that the combination fails to teach the specific type of instruction
being predicted and storing its associated prediction field unpersuasive.
KSR, 550 U.S. at 417 (“If a person of ordinary skill can implement a
predictable variation, § 103 likely bars its patentability.”); see also In re
Rice, 341 F.2d 309, 314 (CCPA 1965) (finding minor differences between
the prior art and a claimed device may be a matter of design choice absent
evidence to the contrary).
5
Appeal 2017-000747
Application 13/449,754
(3) Not clearly shown
Appellants argue Gschwind does not unambiguously teach or suggest
that a bimodal predictor is stored within an instruction. App. Br. 7; Reply
Br. 4—5. More specifically, Appellants argue Gschwind teaches using a
method “to establish a relatively static prediction for the branch, perhaps a
bimodal predictor or hint bits within the branch instruction.” App. Br. 7
(citing Gschwind 117). Appellants argue that “it appears that Gschwind is
describing a bimodal predictor to be one way to establish a static prediction
for a branch, and alternative method is ‘hint bits within the branch
instruction. ’” Id. at 7—8. Appellants further argue that Gschwind does not
teach “how a bimodal predictor is stored within a branch instruction,” nor
“further teach[] of this notion of hint bits.” Id. at 8.
The Examiner finds, and we agree, Gschwind teaches or suggests
encoding within an instruction (i) a bimodal predictor or (ii) hint bits. See
Ans. 14—15 (citing Gschwind 117). More specifically, we agree with the
Examiner that “Gschwind is contemplating several ways to establish a
‘relatively static prediction’ by storing the predictor/hint bits in the branch
itself.” Ans. 15; see also Adv. Act. 2—3. Regardless, encoding (i.e., storing)
hint bits — which in the context of the passage clearly relate to predictions,
such as those that also could be stored “within the pattern history table” —
within the instruction is sufficiently taught or suggested by Gschwind so as
also to teach or suggest the disputed claim language. See Gschwind 117.
We are unpersuaded by Appellants’ arguments that Gschwind’s
teachings are too ambiguous to support this finding. “[I]t is proper to take
into account not only specific teachings of the reference but also the
inferences which one skilled in the art would reasonably be expected to draw
6
Appeal 2017-000747
Application 13/449,754
therefrom.” In re Preda, 401 F.2d 825, 826 (CCPA 1968); In re Paulsen, 30
F.3d 1475, 1480 (Fed. Cir. 1994) (holding that prior art references must be
“considered together with the knowledge of one of ordinary skill in the
pertinent art”).
(4) Combining Grochowski and Gschwind
Appellants argue “there is no motivation to combine Grochowski and
Gschwind as suggested by the Examiner.” App. Br. 9. Appellants argue
Grochowski teaches constructing a history or pattern table for predictions for
a branch instruction — not a producer instruction. App. Br. 9 (citing
Grochowski 113). As to Gschwind, Appellants argue it similarly teaches
history-based prediction for a consumer instruction, and not a producer
instruction. Id. (citing Gschwind 17). Appellants thus contend that “even
if Gschwind’s alleged teaching of encoding a prediction field within a
branch (consumer) instruction is used to prevent aliasing of a branch history
table is applied to Grochowski, Grochowski could at best use this in
preventing aliasing of its own branch history table.” Id.', see also App. Br.
10-11.
The Examiner finds Grochowski and Gschwind are properly
combined. Ans. 15—16; Adv. Act. 2—3. As discussed in more detail above,
the Examiner finds Grochowski teaches or suggests, inter alia, that the
history based branch prediction technique using a history or pattern table can
be applied to predict the evaluation of producer instructions. Ans. 16 (citing
Grochowski 12—13). The Examiner finds “the motivation provided by
Gschwind (to reduce destructive aliasing within a pattern history based
prediction table) is [properly] applied to the history based prediction of the
producer instruction, taught by Grochowski.” Id.', see also Adv. Act. 2—3.
7
Appeal 2017-000747
Application 13/449,754
We find Appellants’ arguments unpersuasive. The Examiner provides
“articulated reasoning with some rational underpinning to support the legal
conclusion of obviousness.” In re Kahn, 441 F.3d 977, 988 (Fed. Cir. 2006).
For example, the Examiner reasons a person of ordinary skill in the art
would have been motivated to combine Grochowski and Gschwind to reduce
destructive aliasing within a history based prediction table. Ans. 16; Adv.
Act. 2—3; see also KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 417 (2007)
(“[I]f a technique has been used to improve one device, and a person of
ordinary skill in the art would recognize that it would improve similar
devices in the same way, using the technique is obvious unless its actual
application is beyond his or her skill.”).
(5) Storing the updated prediction field in memory
Appellants argue the combination of Grochowski and Gschwind fails
to teach or suggest “storing the producer instruction with the updated
prediction field in memory,” as recited in claim 6, and similarly recited in
claims 20 and 30. App. Br. 11—12. More specifically, Appellants argue
“there is no requirement for the updated prediction field to be stored in
memory for this alleged teaching. The comparison and accompanying
update, as alleged, can be performed in entirely combinational logic without
the producer instruction with the updated prediction field in memory.” App.
Br. 12.
The Examiner finds, and we agree, the combination of Grochowski
and Gschwind teaches or suggests the disputed limitation. Ans. 17; Adv.
Act. 3. More specifically, the Examiner finds, and we agree, the
combination teaches or suggests that the prediction values are updated (i.e.,
stored in memory) during execution by the processor which determines if
8
Appeal 2017-000747
Application 13/449,754
the actual values are different from that predicted. Ans. 17 (citing
Grochowski || 10, 21; Gschwind 10, 17); see also In re Preda, 401 F.2d
825, 826 (CCPA 1968) (“[I]t is proper to take into account not only specific
teachings of the reference but also the inferences which one skilled in the art
would reasonably be expected to draw therefrom.”); In re Paulsen, 30 F.3d
1475, 1480 (Fed. Cir. 1994).
CONCLUSION
Based on our findings above, we sustain the Examiner’s rejection of
claims 1, 6, 12, 20, 26, 30, and 34, as well as claims 2—5, 7, 10, 11, 13—16,
18, 19, 21, 24, 25, 27—29, 31, 35, and 36, as Appellants do not provide
separate arguments for their patentability. We also sustain the Examiner’s
rejections of claims 8, 9, 17, 22, 23, 32, and 33, as Appellants rely on their
arguments made with respect to independent claims 1, 12, 26, and 34.
DECISION
We affirm the Examiner’s decision rejecting claims 1—36.
No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a)(l)(iv).
AFFIRMED
9