Ex Parte Archambault et al

Patent Trial and Appeal Board

Jun 10, 2016

12830553 (P.T.A.B. Jun. 10, 2016)

UNITED STA TES p A TENT AND TRADEMARK OFFICE
APPLICATION NO. FILING DATE FIRST NAMED INVENTOR
12/830,553 07/06/2010 Roch G. Archambault
70303 7590 06/14/2016
IBM RALEIGH IPLAW (DL)
C/O DELIZIO LAW, PLLC
15201 MASON ROAD
SUITE 1000-312
CYPRESS, TX 77433
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
ATTORNEY DOCKET NO. CONFIRMATION NO.
CA920090002US 1 1792
EXAMINER
VU, TUAN A
ART UNIT PAPER NUMBER
2193
NOTIFICATION DATE DELIVERY MODE
06/14/2016 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):
USPTO@DELIZIOLA W.COM
USPT02@DELIZIOLA W.COM
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Ex parte ROCH G. ARCHAMBAULT, Y AOQING GAO,
ALLAN R. MARTIN, MARK P. MENDELL,
RAUL E. SIL VERA, and GRAHAM YIN 1
Appeal2014-005285
Application 12/830,553
Technology Center 2100
Before DEBRA K. STEPHENS, JASON V. MORGAN, and
SHARON PENICK, Administrative Patent Judges.
MORGAN, Administrative Patent Judge.
DECISION ON APPEAL
Introduction
This is an appeal under 35 U.S.C. § 134(a) from the Examiner's final
rejection of claims 1-25. We have jurisdiction under 35 U.S.C. § 6(b).
We AFFIRM.
Invention
Appellants disclose instrumentation by a multi-pass profiler of a code
region of an application and execution of the application to identify first
1 Appellants identify International Business Machines Corporation as the
real party in interest. App. Br. 1.
Appeal2014-005285
Application 12/830,553
profile data. Abstract (May 13, 2013). The multi-pass profiler identifies a
delinquent code region and annotates a code sub-region of the delinquent
code region so that re-execution of the application generates second profile
data for the code sub-region. Id.
Exemplary Claim
Claim 1, reproduced below with key limitations emphasized, is
representative:
1. A method comprising:
instrumenting, by a multi-pass profiler, a code region of
an application with first annotations for generating profile data;
executing the application with the first annotations,
wherein executing the application with the first annotations
generates first profile data for the code region;
identifYing, by the multi-pass profiler, from the first
profile data, the code region as a delinquent code region,
wherein the delinquent code region is defined by location of at
least one of the first annotations;
determining, by the multi-pass profiler, second
annotations based, at least in part, on the first profile data and
the at least one of the first annotations that defines the
delinquent code region;
instrumenting, by the multi-pass profiler, based on the
first profile data, a code sub-region of the delinquent code
region with the second annotations for generating profile data,
wherein said instrumenting with the second annotations is finer
grained profiling than said instrumenting with the first
annotations; and
executing the application with the second annotations,
wherein said executing the application with the second
annotations generates second profile data for the code sub-
reg10n.
2
Appeal2014-005285
Application 12/830,553
Rejections
The Examiner rejects claims 1--4, 6-8, 10-18, and 20-24 under 35
U.S.C. § 103(a) as being unpatentable over Kosche et al. (US 2008/0177756
Al; pub. July 24, 2008), Buzbee (US 5,815,720; iss. Sept. 29, 1998), Nair
(US 2004/0054992 Al; publ. Mar. 18, 2004), Wannamaker (US 7,788,657
B2; iss. Aug. 31, 2010), and Patel (US 7,228,404 Bl; iss. July 5, 2007).
Final Act. 2-16.
The Examiner rejects claims 5, 9, 19, and 25 under 35 U.S.C. § 103(a)
as being unpatentable over Kosche, Buzbee, Nair, Wannamaker, Patel,
Burke (US 6,865,730 Bl; iss. Mar. 8, 2005), and Bortnikov et al. (US
6,631,518 Bl; iss. Oct. 7, 2003). Final Act. 16-18.
ISSUE
Did the Examiner err in finding the combination of Kosche, Buzbee,
Nair, Wannamaker, and Patel teaches or suggests: (1) "identifying, by the
multi-pass profiler, ... the code region as a delinquent code region"; (2)
"determining, by the multi-pass profiler, second annotations based ... on the
first profile data and the at least one of the first annotations that defines the
delinquent code region"; and (3) "instrumenting, by the multi-pass profiler,
based on the first profile data, a code sub-region of the delinquent code
region with the second annotations for generating profile data, wherein said
instrumenting with the second annotations is finer grained profiling than said
instrumenting with the first annotations," as recited in claim 1?
ANALYSIS
In rejecting claim 1, the Examiner finds, inter alia, that Kosche's
analysis engine probing of an event space to identify performance
bottlenecks teaches or suggests identifYing a code region as a delinquent
3
Appeal2014-005285
Application 12/830,553
code region. Final Act. 3 (citing Kosche iii! 69, 89, 93, 112, 242, and 312,
Figs. 5, 24). The Examiner finds Kosche teaches or suggests instrumenting
code regions and executing an application, thus profiling the code regions.
Final Act. 3. The Examiner further finds Kosche teaches or suggests re-
instrumenting a delinquent code region based on results from code regions
profiling (determining second annotations). Id. at 5. However, The
Examiner relies on other references, in combination with Kosche, to teach or
suggest determining, by the multi-pass profiler, second annotations based
on the first profile data and at least one of the first annotations (id. at 4) and
instrumenting, by the multi-pass profiler, based on the first profile data, a
code sub-region of the delinquent code region with the second annotations
for generating profile data, wherein said instrumenting with the second
annotations is finer grained profiling than said instrumenting with the first
annotations (id. at 8). In particular, the Examiner finds such determining
and instrumentation is taught or suggested by Buzbee' s teaching of a
compiler that adds new code annotations to query more detailed run-time
behavior when profile information from a previous compilation and
execution is available. Id. at 5---6 (citing Buzbee, Figs. 5---6, col. 5, 1. 56-col.
6, 1. 28); see also Final Act. 7; Ans. 21-22.
Appellants contend the Examiner erred because Kosche' s "analysis
engine does not identify the performance bottlenecks itself." App. Br. 10.
"Instead, 'the use of the analysis engine and graphical user interface thereof
may allow a user to identify data motion bottlenecks."' Id. (citing Kosche
if 69). However, we find the analysis of Kosche "may include displaying
costs associated with various hardware and software components of the
system and the application being profiled." Kosche if 69. Thus, we find the
4
Appeal2014-005285
Application 12/830,553
analysis engine of Kosche produces the details as to which software
components or code regions have high-costs (i.e., are delinquent) and
accordingly teaches or suggests identifying a code region as a delinquent
code region.
Appellants further contend the Examiner erred because Kosche
merely teaches a "process of profiling an application, optimizing the
application based on the generated profile data, then profiling the application
again" and does not teach determining second annotations based on
previously generated profile data. App. Br. 10 (citing Final Act. 4).
Appellants argue Kosche only teaches "eliminating one set of bottlenecks,
allowing future instrumentation (and thus profiling) to help identify a second
set of bottlenecks." App. Br. 10. However, the Examiner relies on Kosche
in combination with Buzbee to teach or suggest the second instrumenting.
See Final Act. 5-7. Moreover, the claims do not preclude optimization
efforts between the two instrumenting steps.
Appellants next contend Buzbee does not cure the alleged deficiencies
of Kosche because Buzbee also "merely discloses profiling an application,
optimizing the application, and profiling again." App. Br. 13. In particular,
although Buzbee teaches a process that includes "prompt[ing] the addition of
new code annotations to query more detailed run-time behavior" (Buzbee
col. 5, 11. 64--65), Appellants argue
it is not clear what "more detailed run-time behavior" means.
For example, querying "more detailed run-time behavior" very
well could mean simply generating more statistical data (i.e.,
instead of just counting cache misses, counting each memory
reference). Or, for example, Buzbee could be describing
profiling a subset of the application (say, only half of the
memory references) first, then profiling a larger subset (all of
5
Appeal2014-005285
Application 12/830,553
the memory references) on the second pass. Such a scenario
provides more detailed information, but increases the number of
annotations and does not make the profiling finer-grained. In
fact, Buzbee could be describing as increasing the granularity
on the second pass, as opposed to reducing it. For example, the
first profiling pass might include only instruction level data,
whereas the second profiling pass might also include function
level data, thus increasing the granularity while providing more
detailed data.
App. Br. 15-16; see also Reply Br. 2-3.
Appellants' arguments are unpersuasive because, as the Examiner
correctly finds, Buzbee' s feedback illustrates a "scenario wherein at least
one profiling/code annotating pass is followed by another profiling pass,
using ... the profile data . . . from one instrumentation (first pass) instance
to reconfigure (with a refining effect) the subsequent code instrumentation
(second pass)." Ans. 22. This instrumentation refinement is depicted in
Buzbee's Figure 5, reproduced below:
I SOURCE
CODE
41
COMPILER:
ADD CODE ANNOTATIONS
TO QUERY RUN-TIME
BEHAVIOR OF THE
APPLICATION. WHEN
PROFILE INFO IS
AVAILABLE, USE IT TO AID
OPTIMIZATION AND
PROMPT ADDITION OF NEW
CODE ANNOTATIONS TO
QUERY MORE DETAILED
RUN-TIME BEHAVIOR
42
FIGURE 5 PROFILE
INFORMATION 45--------
6
DYNAMIC TRANSLATOR:
RUN UNDER DYNAMIC
TRANSLATION, USING
ANNOTATIONS TO
DETERMINE WHICH RUN-
TIME PROFILE
INFORMATION TO
COLLECT
44
Appeal2014-005285
Application 12/830,553
We find Buzbee's Figure 5 illustrates compiler 42 using profile
information 45, produced through a first pass run of the compiled and
optimized application, to "prompt addition of new code annotations to query
more detailed run-time behavior" during a second run compilation of source
code 41 for repeat optimization (step 43) and execution (step 44). See also
Buzbee col 5, 11. 45-55.
This prompted addition of new code annotations to query more
detailed run-time behavior teaches or suggests the claimed finer grained
profiling because the "annotations are adjusted on each compilation to
obtain additional profile information which will be used in future
compilations." Buzbee col. 6, 11. 22-24 (emphasis added). That is, the more
detailed run-time behavior information obtained through Buzbee provides
additional, finer-grained information regarding the compiled application.
Appellants' arguments that Buzbee' s process of querying more
detailed run-time behavior might merely generate more statistical data or
(e.g., memory references rather than just cache misses) or an expanded
subset of the application (e.g., all references rather than half the references)
(App. Br. 15) are unpersuasive because Appellants' do not distinguish the
claimed finer grained profiling from these proffered hypotheticals. In both
examples, information that was ignored in a first pass is queried in a second
pass, thus providing for a more detailed, finer grained profile of the
instrumented application.
Appellants' argument that a first profile pass might include
instruction-level data, but a second pass might merely add function-level
data (id. at 15-16), is also unpersuasive. The proffered hypothetical would
either result in previously ignored information being queried or would fail to
7
Appeal2014-005285
Application 12/830,553
generate additional profile information in the manner taught or suggested by
Buzbee. That is, if only a subset of a profiled function's instructions had
previously been profiled, then adding functional-level data would obtain
data for the subset of the profiled function's instructions that had not been
profiled. Alternatively, if all of a profiled function's instructions had
previously been profiled, then the functional-level data would be redundant
to the aggregate instruction-level data, contrary to Buzbee' s teaching of
adding new code annotations to query more detailed run-time behavior. See
Buzbee col. 5, 11. 64---65.
Appellants further argue that claim 1 "specifically recites that all
operations, except for the actual execution of the application, are performed
by the multi-pass profiler," rather than through "any form of developer's
review of the profile results to determine the second annotations." App. Br.
20. However, Buzbee specifically teaches placement of annotations into
object code "by a compiler." Buzbee col. 2, 11. 24--25. This is in contrast to
user-added annotations, which Buzbee also teaches. See Buzbee, Fig. 6
(step 56); cf id., Fig. 5. Thus, Buzbee, in combination with Kosche, teaches
or suggests a multi-pass profile (e.g., a compiler having profiling features)
that performs the claimed determining and second-pass instrumenting steps.
Moreover, Appellants' arguments are incommensurate with the scope of the
claimed invention, which does not preclude involvement by a user or
developer as part of or in addition to the claimed steps.
Appellants further argue that the claimed invention is directed to
"using multiple instrumentation cycles to identifY the bottlenecks in the first
place" and that "performing the same instrumentation process on different
program code to identify different bottlenecks in each version of program
8
Appeal2014-005285
Application 12/830,553
code cannot be said to render obvious performing different instrumentation
processes on the same code to identify the bottlenecks in the first place."
App. Br. 10-11; see also Reply Br. 6. However, the claimed invention does
not preclude optimization of an application between passes. Moreover,
Buzbee teaches or suggest multi-pass compilation (with instrumentation) in
which source code 41 does not need to change for more detailed run-time
behavior (queried when profile information 45 exists) to be queried. See
Buzbee, Fig. 5.
Appellants provide additional arguments with respect to the
Examiner's citation to Nair, Wannamaker, and Patel. See, e.g., App. Br. 11-
12, 16-19; Reply Br. 2-3. However, the Examiner's findings and reasoning
show that the combination of Kosche and Buzbee alone suffices to show that
the disputed recitations would have been obvious. This makes Appellants'
additional arguments moot and thus unpersuasive. Therefore, we agree with
the Examiner that the combination of Kosche, Buzbee, Nair, Wannamaker,
and Patel teaches or suggests: (1) "identifying, by the multi-pass profiler, ...
the code region as a delinquent code region"; (2) "determining, by the multi-
pass profiler, second annotations based ... on the first profile data and the at
least one of the first annotations that defines the delinquent code region";
and (3) "instrumenting, by the multi-pass profiler, based on the first profile
data, a code sub-region of the delinquent code region with the second
annotations for generating profile data, wherein said instrumenting with the
second annotations is finer grained profiling than said instrumenting with the
first annotations," as recited in claim 1.
Accordingly, we sustain the Examiner's 35 U.S.C. § 103(a) rejection
of claim 1, and claims 2--4, 6-8, 10-18, and 20-24, which Appellants do not
9
Appeal2014-005285
Application 12/830,553
argue separately. App. Br. 21-22. Appellants argue the remaining claims
are patentable for similar reasons. Id. Therefore, for the reasons above, we
also sustain the Examiner's 35 U.S.C. § 103(a) rejection of claims 5, 9, 19,
and 25.
DECISION
We affirm the Examiner's decision rejecting claims 1-25.
No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a). See 37 C.F.R.
§ 41.50(±).
AFFIRMED
10