`
`
`UNITED STATES AND TRADEMARK OFFICE
`
`––––––––––––––––––
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`––––––––––––––––––
`
`MICROSOFT CORPORATION,
`Petitioner,
`
`v.
`
`DIRECTSTREAM, LLC,
`Patent Owner.
`
`––––––––––––––––––
`
`Case No. IPR2018-01605, -01606 and -01607
`U.S. Patent No. 7,620,800 B2
`
`––––––––––––––––––
`
`PETITIONER’S REPLY TO PATENT OWNER’S RESPONSE
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`
`TABLE OF CONTENTS
`
`
`
`INTRODUCTION ......................................................................................... 1
`I.
`II. DIRECTSTREAM IMPROPERLY INCORPORATES BY
`REFERENCE ................................................................................................. 1
`III. DIRECTSTREAM MISCHARACTERIZES THE TECHNICAL
`BACKGROUND OF THE 800 PATENT ................................................... 2
`IV. DR. STONE’S TESTIMONY IS DETAILED AND RELIABLE............. 3
`A. Dr. Stone Addressed The Inventions Described and Claimed in The
`800 Patent .............................................................................................. 3
`B. Dr. Stone Based His Analysis On The Prior Art, Not Hindsight .......... 5
`C. The Petition and Dr. Stone Provided Detail Bases and Factual
`Underpinnings For the Obviousness Analysis ...................................... 8
`D. Splash2 Instantiated Looping In Its Reconfigurable Hardware .......... 15
`V. DIRECTSTREAM’S CLAIM CONSTRUCTIONS SHOULD BE
`REJECTED .................................................................................................. 19
`A. The Board Correctly Construed “Pass Computed Data Seamlessly” 20
`B. The Board Correctly Construed “Systolic” ......................................... 26
`C. The Board Correctly Construed “Stream Communication” ............... 26
`D. The Board Correctly Construed “Computational Loop” .................... 31
`VI. THE CHALLENGED CLAIMS ARE UNPATENTABLE ..................... 33
`A. The Challenged Claims are Unpatentable over Splash2 ..................... 33
`1.
`Splash2 Discloses Stream Communication .............................. 33
`2.
`Splash2 Discloses The Claimed Computational Loops ............ 34
`3.
`RaPiD Discloses The Claimed Computational Loops .............. 36
`4.
`The Prior Art Unambiguously Discloses Seamlessly Passing
`Computed Data Between Computational Loops ...................... 40
`B. DirectStream Fails to Overcome The Obvious Combinations Set Forth
`in The Petition ..................................................................................... 45
`1.
`The Combination of Splash2 and Gaudiot Would Have Been
`Obvious ..................................................................................... 47
`
`i
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`
`2.
`
`The Combination of Splash2 and Jeong Would Have Been
`Obvious ..................................................................................... 48
`C. The Objective Indicia in this Case Indicate Nonobviousness ............. 48
`VII. CONCLUSION ............................................................................................ 51
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`ii
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`
`TABLE OF AUTHORITIES
`
`
`
` Page(s)
`
`Cases
`Amgen Inc. v. Hoechst Marion Roussel, Inc.,
`314 F.3d 1313 (Fed. Cir. 2003) .......................................................................... 12
`Apple Inc., v. Contentguard Holdings, Inc.,
`No. IPR2015-00445, 2015 WL 4264955 (PTAB July 9, 2015) ......................... 36
`Caterpillar v. Wirtgen,
`IPR2017-02186, Paper 10 (PTAB May 1, 2019) ..................................... 6, 10, 11
`Cisco Sys., Inc. v. C-Cation Techs., LLC,
`IPR2014-00454, Paper 12 (PTAB Aug. 29, 2014) ............................................... 2
`Custom Accessories v. Jeffrey-Allan Indus.,
`807 F. 2d 955 (Fed. Cir. 1986) ................................................................... 8, 9, 11
`Icon Health & Fitness, Inc. v. Strava, Inc.,
`849 F.3d 1034 (Fed. Cir. 2017) .......................................................................... 55
`Intercontinental Great Brands LLC v. Kellogg N. Am. Co.,
`869 F.3d 1336 (Fed. Cir. 2017) .................................................................... 56, 57
`Ex Parte Jud,
`Appeal 2006-001061 (BPAI Jan. 20, 2007) ......................................................... 8
`KSR Int’l Co. v. Teleflex, Inc.,
`550 U.S. 398 (2007) .............................................................................................. 7
`Limelight Networks Inc. v. Mass. Inst. Tech.,
`IPR2017-00249, Paper 9, (PTAB May 18, 2017) ........................................ 2, 5, 6
`In re McLaughlin,
`443 F.2d 1392 (CCPA 1971) ................................................................................ 6
`In re Merck & Co., Inc.,
`800 F.2d 1091 (Fed. Cir. 1986) .............................................................. 41, 51, 52
`
`iii
`
`
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`Phillips v. AWH Corp.,
`415 F.3d 1303 (Fed. Cir. 2005) .................................................................... 20, 36
`Pitney Bowes, Inc. v. Hewlett-Packard Co.,
`182 F.3d 1298 (Fed. Cir. 1999) .......................................................................... 34
`Polaris Indus., Inc. v. Arctic Cat, Inc.,
`882 F.3d 1056 (Fed. Cir. 2018) .......................................................................... 56
`Regents of Univ. of New Mexico v. Knight,
`321 F.3d 1111 (Fed. Cir. 2003) .......................................................................... 53
`Robocast, Inc., v. Apple Inc.,
`39 F.Supp.3d 552 (D. Del. 2014) ........................................................................ 12
`Securus Techs. Inc. v. Glob. Tel*Link Corp.,
`701 F. App’x 971 (Fed. Cir. 2017) ..................................................................... 30
`Specialty Composites v. Cabot Corp.,
`845 F.2d 981 (Fed. Cir. 1988) ............................................................................ 23
`VirnetX, Inc. et al. v. Apple Inc.,
` No. 2019-01050, (Fed. Cir., Nov. 23, 2019). ..................................................... 33
`
`Other Authorities
`37 C.F.R. § 42.6(e) ................................................................................................... 65
`37 C.F.R. § 42.24 ............................................................................................... 63, 64
`
`
`iv
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`I.
`
`INTRODUCTION
`
`
`
`DirectStream’s arguments are utterly disconnected from the language of the
`
`claims, the analysis of the Petitions, and the record in these proceedings. It asks
`
`that numerous extraneous limitations be added to its claims, without even
`
`attempting to justify them as reflecting the broadest reasonable interpretation, or
`
`the ordinary meaning of its claim language. It makes numerous unexplained
`
`arguments, without citation to the record or citing to portions of the record that
`
`bear no relationship to its argument. It throws various baseless accusations against
`
`Dr. Stone, citing to testimony that does not support its point and in many cases
`
`says exactly the opposite. Its scattershot arguments, however, cannot avoid the
`
`fact that the challenged claims read squarely on the prior art. They are therefore
`
`unpatentable, and the Board should so find.
`
`II. DIRECTSTREAM IMPROPERLY INCORPORATES BY
`REFERENCE
`
`DirectStream cites to large portions of the record spanning numerous pages
`
`and paragraphs without analysis, discussion or pinpoint citation. For example, in
`
`one instance, DirectStream cites to over 5300 pages of information without
`
`specific page cites or explanation, including 295 paragraphs of its expert
`
`declarations. IPR2018-01605 Patent Owner Response (“Response”), 19 n.4. In
`
`another, it cites to 194 paragraphs of expert testimony. Response, 29. It employs
`
`1
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`this technique throughout its brief, as demonstrated in the chart we include as
`
`
`
`Appendix A detailing over two dozen such instances.
`
`Such citations constitute improper incorporation by reference. Cisco Sys.,
`
`Inc. v. C-Cation Techs., LLC, IPR2014-00454, Paper 12, 7–10 (PTAB Aug. 29,
`
`2014). The Board should not consider any of the information set forth in the
`
`citations identified in Appendix A. Id.; Limelight Networks Inc. v. Mass. Inst.
`
`Tech., IPR2017-00249, Paper 9, 7-8 (PTAB May 18, 2017).
`
`III. DIRECTSTREAM MISCHARACTERIZES THE TECHNICAL
`BACKGROUND OF THE 800 PATENT
`
`DirectStream spends the first 17 pages of its response describing various
`
`aspects of what it calls “high performance computing,” or “HPC,” and the use of
`
`FPGAs. Response, 1-17. The challenged claims of the 800 Patent are not limited
`
`to “high performance computing” and that phrase does not appear in the intrinsic
`
`record. Nor has DirectStream ever requested that any claim language be construed
`
`to require “high performance.”
`
`Instead, the 800 Patent expressly states that “[t]he present invention relates,
`
`in general, to the field of computing systems and techniques. More particularly, the
`
`present invention relates to multi-adaptive processing systems and techniques for
`
`enhancing parallelism and performance of computational functions.” EX1005,
`
`1:39-43. The patent describes “multi-adaptive processor elements” as
`
`2
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`reconfigurable processors, EX1005, 2:8-9, but never as “high performance
`
`
`
`computers.”
`
`In any event, the Splash 2 computer system relied on in the Petition
`
`(EX1007, “Splash2”) was a “high performance computer” and was reconfigurable
`
`through its use of FPGAs. For example, Splash2 explains that for the particular
`
`genetic sequence comparison application described in Chapter 8 and relied on in
`
`the Petition, the Splash 2 system provided performance that was several orders of
`
`magnitude higher than a number of high performance computers available at the
`
`time. See EX1007, 107-08. Thus, to the extent the use of FPGAs in high
`
`performance computing has any relevance to the issues to be decided here, it was
`
`the developers of the Splash 2 system who first achieved that milestone, not the
`
`named inventors of the 800 Patent.
`
`IV. DR. STONE’S TESTIMONY IS DETAILED AND RELIABLE
`DirectStream next resorts to unsubstantiated attacks on Petitioner’s expert,
`
`Dr. Harold Stone, but it never analyzes any specific opinions of his it thinks are not
`
`credible and its citations do not support its attacks.
`
`A. Dr. Stone Addressed The Inventions Described and Claimed in
`The 800 Patent
`
`DirectStream contends, for example, that “Dr. Stone fundamentally failed to
`
`understand what the problems the patents at issue are teaching,” Response, 18, but
`
`3
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`its citations to the record do not support that assertion and it never explains how
`
`
`
`they might. The only cited document that even relates to Dr. Stone’s opinions—
`
`EX2065—is a copy of Dr. Stone’s deposition transcript from the IPR2018-01594
`
`proceedings, not this one. The cited portions, moreover, relate to a person of skill
`
`in the art in that proceeding, and to multiprocessors, which are not recited in the
`
`claims of the 800 Patent. Indeed, the cited portions of EX2065 do not discuss any
`
`“problems” to be solved by any of the DirectStream patents, let alone the 800
`
`Patent. The other references cited by DirectStream relate to unrelated technology
`
`or background materials in the art.
`
`DirectStream further contends that “to arrive at his conclusions, Dr. Stone
`
`ignored all of the problems and issues of microprocessors, ASIC chips, FPGA
`
`capabilities to perform floating point operations, high-performance computing
`
`design challenges, or web search algorithms.” Response, 18. However, the claims
`
`of the 800 Patent do not recite “microprocessors, ASIC chips, FPGA capabilities to
`
`perform floating point operations, high-performance computing design challenges,
`
`or web search algorithms,” and none of those issues have anything to do with
`
`whether the claims read on the prior art. Moreover, here again, DirectStream relies
`
`on Dr. Stone’s deposition in another IPR proceeding, and cites portions, without
`
`explanation of their relevance, that do not support DirectStream’s contentions or
`
`relate to any of the issues the Board needs to decide. See id.
`4
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`
`B. Dr. Stone Based His Analysis On The Prior Art, Not Hindsight
`DirectStream asserts that Dr. Stone’s analysis is “grounded in hindsight
`
`
`
`bias,” Response, 19, 115-16, but that is demonstrably untrue, as Dr. Stone’s
`
`analysis is supported by numerous citations to evidence in the prior art.1 Such
`
`prior art based analysis precludes any finding that improper hindsight was
`
`employed. See Caterpillar v. Wirtgen, IPR2017-02186, Paper 10 at 26 (no
`
`hindsight where analysis is based on “substantive evidentiary underpinnings from
`
`the prior art as well as the knowledge and level of expertise of a person of ordinary
`
`skill in the art.”); see also In re McLaughlin, 443 F.2d 1392, 1395 (CCPA 1971).
`
`DirectStream further contends that,
`
`
`
`1 See, e.g., EX1003¶123 (citing EX1007, published 1996); id.,¶126(citing EX1041,
`published 2000, EX1043, published 1997);; id,¶160 (citing EX1035, published
`1999); id., ¶161 (citing EX1008, published 1985); id., 96-97 (citing EX1010,
`published 1987); id., ¶165 (citing EX1021, published 1982); id., ¶251(citing
`EX1012, published in 1998); id., ¶¶288, 486 (citing EX1053, published 1990 and
`EX1057, published 2001); id., ¶302 (citing EX1009, published 1997); id., ¶335
`(citing EX1049, published 1996 and EX1050, published 2001); id., ¶344 (citing
`EX1011, published 1997); id., ¶352 (citing EX1051, published 1996); id., ¶381
`(citing EX1052, published 1999 and EX1058, published 1998); id., ¶393 (citing
`EX1061, published 1997); id., ¶428 (citing EX1062, issued 1992 and EX1063,
`published 1993).
`
`5
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`
`Petitioner uses its experts to define the POSITA by hindsight
`reference to the Patentee’s solutions using reconfigurable components,
`such as FPGAs … rather than by reference to the problem the patent
`was trying to solve.... i.e., various applications in accelerating web
`search functionality or increasing computing times in high-
`performance computing systems using reconfigurable components as
`part of the solution.
`
`
`
`Response, 20. The 800 Patent, however, never defines the problem it was
`
`attempting to solve as “accelerating web search functionality” or as limited to
`
`“high-performance computing systems.” See EX1005, 1:39-3:9.
`
`Moreover, the challenged claims of the 800 Patent recite “[a] method for
`
`data processing in a reconfigurable computing system,” e.g., EX1005, 12:45-46, so
`
`defining the level of skill in terms of reconfigurable computing is entirely proper.
`
`Indeed, it would have been legally erroneous to focus only on “the problem the
`
`patent was trying to solve,” as DirectStream contends. KSR Int’l Co. v. Teleflex,
`
`Inc., 550 U.S. 398, 420 (2007) (“The first error of the Court of Appeals in this case
`
`was to foreclose this reasoning by holding that courts and patent examiners should
`
`look only to the problem the patentee was trying to solve.”)
`
`It is also well-settled that defining the level of ordinary skill based on what a
`
`Skilled Artisan would have known and known how to do is entirely proper: “At its
`
`core, however, the question of the level of ordinary skill in the pertinent art boils
`
`6
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`down to a question of what the hypothetical person with such skill would have
`
`
`
`known (and known how to do).” Ex Parte Jud, Appeal 2006-001061 at 3 (BPAI
`
`Jan. 20, 2007) (precedential opinion) (emphasis added).
`
`A Skilled Artisan is presumed to be aware of the content of the prior art,
`
`Custom Accessories v. Jeffrey-Allan Indus., 807 F. 2d 955, 962 (Fed. Cir. 1986),
`
`and it is undisputed that the prior art to the 800 Patent included knowledge “about
`
`the programming, design and operation of computer systems based on
`
`reconfigurable components,” including the use of hardware description languages,
`
`such as VHDL, to configure FPGAs. EX1007, 36; EX1078 65:19-66:3; EX1075
`
`36:6-20; EX1079, viii-xi; 8-16. Even DirectStream’s expert previously testified
`
`that he believed “a person having ordinary skill in the art (“PHOSITA”) must be
`
`experienced in developing with … hardware description languages, and the
`
`unique problems involved with programming FPGAs and FPGA based systems.”
`
`EX2029, 6 (emphasis added).
`
`Given the undisputed record evidence that how to program and design a
`
`reconfigurable computer system was known in the prior art, DirectStream’s
`
`suggestion that a Skilled Artisan would not be knowledgeable of such techniques
`
`must be rejected as legally improper. See Ex Parte Jud, Appeal 2006-001061 at 6-
`
`7. Dr. Stone’s characterization of the Skilled Artisan was therefore fully
`
`compliant with the law and supported by the record.
`7
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`
`DirectStream’s expert, on the other hand, bases his opinions in this
`
`
`
`proceeding on a legally erroneous understanding of the law. As he conceded on
`
`cross-examination, he believed that a Skilled Artisan was only presumed to be
`
`aware of what the expert called the “prior relevant art.” EX1075, 27:4-12. That,
`
`of course, is not the law, Custom Accessories, supra, and led the expert into his
`
`mistaken conclusion that a Skilled Artisan would not have been aware of such
`
`basic prior art techniques as how to program an FPGA. The expert’s testimony
`
`should be given no weight.
`
`C. The Petition and Dr. Stone Provided Detailed Bases and Factual
`Underpinnings
`
`DirectStream also argues that “Petitioner provides no factual basis or
`
`rational underpinning for its suggested combinations of prior art references.
`
`Instead, Dr. Stone’s testimony regarding the POSITA motivation to combine is
`
`conclusory, consisting of nothing more than his own ipse dixit,” citing only to a
`
`deposition transcript from the -01594 proceeding. Response, 20. Similarly,
`
`DirectStream asserts that “Petitioner fails to provide any support on how such an
`
`artisan would combine the references cited.” Response, 23. The only support
`
`DirectStream cites for these accusations, however, are “’1594 Petition at passim”
`
`and “EX1003¶¶passim”. Response, 22.
`
`8
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`
`These assertions are baseless. The motivation to combine analysis in Dr.
`
`
`
`Stone’s declaration alone spans 44 paragraphs, contains 4811 words and cites to 16
`
`references in the prior art. See EX1003¶¶168-73, 282-88, 331-39, 377-85, 424-29.
`
`For each cited combination, Dr. Stone identifies multiple reasons to combine,
`
`EX1003¶¶171-73, 286-88, 334-39, 480-85, 427-29, and explains exactly how each
`
`combination would be carried out, EX1003¶¶167, 261, 264-281, 315-30, 365-378,
`
`409-423, 461, 464, 467-79. That is not an “ipse dixit” analysis, and Dr. Stone’s
`
`broad-based reliance on evidence from the prior art precludes any conclusion that
`
`improper hindsight is being used. See Caterpillar, IPR2017-02186, Paper 10 at 26.
`
`Indeed, in its response DirectStream never even addresses, and therefore
`
`does not dispute, the specific motivations to combine identified in the Petition and
`
`supported by Dr. Stone’s testimony and the evidence he cites. Response, 18-29.
`
`Instead, DirectStream cites only to Dr. Stone’s deposition testimony in the -01594
`
`proceeding, which did not relate to any motivations or evidence at issue in this
`
`proceeding. See Response, 20, citing EX2065.
`
`DirectStream also contends that Dr. Stone fails to note “the inefficiencies
`
`and challenges in designing multiprocessor systems and the component parts
`
`needed to perform HPC computations at the time of the invention.” Response, 21.
`
`It argues that Dr. Stone failed to address some of the design considerations
`
`described in his 1987 textbook. Id. at 21-22. It further states that “Dr. Stone
`9
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`
`testified he assumed his use of prior art in various combinations would be enabling
`
`without explaining how a POSITA would actually make that combination,” once
`
`again citing testimony in the -01594 proceeding, not this proceeding. Id., 24. None
`
`of that is accurate.
`
`The claims at issue here do not require “multiprocessor systems and the
`
`component parts needed to perform HPC computations,” so Dr. Stone’s supposed
`
`failure to address those issues is irrelevant. Similarly, none of the knowledge or
`
`design considerations Dr. Stone supposedly failed to discuss have anything to do
`
`with the issues to be decided here, and DirectStream does not even attempt to
`
`argue that they do.
`
`Moreover, a Skilled Artisan is presumed to be aware of the content of the
`
`prior art, including the design considerations set forth in Dr. Stone’s 1987
`
`textbook. Custom Accessories, 807 F. 2d at 962. Such a person would therefore
`
`have been able to apply those considerations to the combinations analyzed in the
`
`Petitions, and DirectStream never argues otherwise. There was, therefore, no
`
`reason for Dr. Stone to analyze them in his declaration.
`
`DirectStream’s assertions regarding enablement are similarly baseless. Prior
`
`art patents and publications are presumptively enabling, Amgen Inc. v. Hoechst
`
`Marion Roussel, Inc., 314 F.3d 1313, 1355 (Fed. Cir. 2003); Robocast, Inc., v.
`
`10
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`Apple Inc., 39 F. Supp .3d 552, 565 (D. Del. 2014), and DirectStream makes no
`
`
`
`attempt to rebut that presumption by arguing a lack of enablement.
`
`Moreover, for each cited combination Dr. Stone explained how the
`
`combination would be made. EX1003¶¶167, 261, 264-81, 315-30, 365-78, 409-
`
`23, 461, 464, 467-79. He also testified that “[i]t is my opinion that none of the
`
`combinations would require undue experimentation,” EX2063, 223:18-25, and that
`
`each combination could be “readily” made in the sense that a Skilled Artisan could
`
`make the combinations without undue experimentation, id., 223:7-17;
`
`EX1003¶¶160, 190, 285, 333, 339, 379, 426, 483. Neither DirectStream nor its
`
`expert contests those opinions.
`
`In his obviousness analysis, moreover, Dr. Stone analyzed, with copious
`
`citations to the prior art, numerous advantages that would result from the obvious
`
`modifications and combinations he proposes, including:
`
`• the enhanced performance and efficiencies described in Splash2,
`
`EX1003¶¶158-61, 188-90,
`
`• the scheduling efficiencies, scalability, and flexibility of data driven
`
`systems, such as that of Gaudiot, EX1003¶¶172-73,
`
`• the prior art statements acclaiming Splash2 as a “very successful
`
`example” of a reconfigurable computer system and the advantageous
`
`11
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`
`architectural properties of Splash2 for image processing applications,
`
`
`
`such as disclosed in RaPiD, EX1003¶¶334-39,
`
`• the explicit mapping of systolic structures to the system of Jeong,
`
`EX1003¶¶412-16, 424-29,
`
`• the popularity of, and ready combinability with Splash2, of automatic
`
`target recognition systems, such as disclosed in Chunky SLD
`
`EX1003¶¶377-85,
`
`• the supercomputer performance on molecular biology applications
`
`already achieved by Splash2 and popularity of parallel computing
`
`systems in such applications, such as disclosed in Roccatano,
`
`EX1003¶¶282-288.
`
`DirectStream never even addresses these advantages, let alone attempts to
`
`rebut them or show countervailing disadvantages. With its silence, it concedes Dr.
`
`Stone’s analysis is accurate.
`
`DirectStream also contends, without citation to any particular portion of Dr.
`
`Stone’s 277-page declaration, that Dr. Stone failed to disclose his underlying facts
`
`and data. Response, 22-29. DirectStream never identifies any particular issue or
`
`claim element for which insufficient facts or data are provided; nor could it. The
`
`facts and data relevant to a patentability analysis are, of course, the claims as
`
`properly construed, the prior art, and any additional evidence of level of skill,
`12
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`motivation to combine, etc. Dr. Stone provides that information in copious detail.
`
`
`
`His analysis of claim 1 is by itself 62 pages long with numerous pinpoint cites to
`
`the prior art. DirectStream’s hand waving and unsupported assertions to the
`
`contrary cannot erase that detailed analysis from the record.
`
`DirectStream also includes a footnote citing testimony supposedly showing
`
`that Dr. Stone “merely assumed all the benefits and ignored any of the drawbacks”
`
`of the technology. Response, 26 n.6. The footnote cites only to deposition
`
`testimony taken in connection with the -01594 proceeding, not this one, and the
`
`cited portions of that transcript show nothing of the sort. For example:
`
`• At 65:17-68:22, Dr. Stone noted that he generally focuses on the
`
`positives when considering a combination of prior art references,
`
`because “you don't know what you can't achieve.” EX2065,
`
`67:10. However, he also noted that disadvantages would be
`
`considered, if they were relevant to the combination. Id., 67:4-13.
`
`• At 142:22-144:10, while Dr. Stone agreed that his declaration did not
`
`address a specific example of a disadvantage of multiprocessors,
`
`multiprocessor are not recited in the claims of the 800 Patent.
`
`• At 145:9-146:25, Dr. Stone notes that he “considered the advantages
`
`and disadvantages in forming the combinations,” EX2065, 145:16-
`
`13
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`
`17, and proceeds to specifically explain how the disadvantages
`
`
`
`factored into his analysis. Id., 145:17-146:5.
`
`• At 161:13-162:25, Dr. Stone correctly states that the legal standard for
`
`obviousness “says that the person of ordinary skill in the art should be
`
`motivated, should try to be -- understand that the combinations can be
`
`made and that they would -- there would be some predictable
`
`results” EX2065, 161:20-25, and that this standard does not require
`
`the “best solution”—“just that it has to be obvious and it will -- it will
`
`solve a problem.” Id., 162:2-6.
`
`• At 187:9-188:11, Dr. Stone indicates that he would need to investigate
`
`whether certain “potential drawbacks of FPGAs in 2008” (emphasis
`
`added) would have been drawbacks that existed in 2001, but
`
`DirectStream cites to no such “potential drawbacks” from 2008 that
`
`could affect Dr. Stone’s analysis.
`
`• At 203:14-207:13, Dr. Stone gives examples of disadvantages
`
`discussed in certain prior art references that he considered in his
`
`analysis. EX2065, 205:19-206:5.
`
`Thus, contrary to DirectStream’s unsupported assertions, Dr. Stone’s
`
`testimony consistently articulates fact based, rational reasons to combine the
`
`various cited references, in which he considers both the advantages and
`14
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`disadvantages of the prior art. Moreover, DirectStream neither disputes Dr.
`
`
`
`Stone’s testimony on these points nor identifies any specific disadvantage relevant
`
`to the patentability of the claims that Dr. Stone may not have considered.
`
`At various points, DirectStream also cites to various technical issues relating
`
`to reconfigurable computing, such as “the issues of reconfigurable programming,
`
`processor speed, FPGA speed, benchmarking, bottlenecking, and cost/benefit
`
`analysis of overhead introduction as applied to HPC applications of reconfigurable
`
`programming, processor speed, FPGA speed, benchmarking, bottlenecking, and
`
`cost/benefit analysis of overhead introduction as applied to HPC applications.”
`
`Response, 30. But DirectStream never explains what any of those things have to
`
`do with the challenged claims, or how they might relate to how those claims read
`
`on Splash2 or the various prior art combinations set forth in the Petitions. Indeed,
`
`the Splash 2 system already included FPGAs, so any technical hurdles that may
`
`have existed when implementing a reconfigurable computing system were already
`
`overcome by the inventors of that system. DirectStream’s bare mention of them in
`
`its brief is thus irrelevant.
`
`Splash2 Formed Looping In Its Reconfigurable Hardware
`
`D.
`DirectStream also suggests that the Splash 2 system was incapable of
`
`performing loops in the reconfigurable hardware, confusingly citing to
`
`“§IV.B.2.b.” Response, 26. The Patent Owner Response, however, does not
`
`15
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`include a “§IV.B.2.b,” and §IV relates only to a person of ordinary skill; not to
`
`
`
`Splash2.2
`
`DirectStream makes this argument at several other points in its brief, see
`
`Response, 28, 83-85, 107, 115, 120, but it provides no more cogent analysis. In
`
`particular, for this theory DirectStream relies on one paragraph of the doctoral
`
`thesis of Richard Halverson, EX2169. See, e.g., Response, 83. Dr. Halverson had
`
`nothing to do with the development of the Splash 2 system and the single
`
`paragraph in his dissertation does not even purport to be a detailed or fulsome
`
`description of that machine. Nor does that paragraph have anything to do with the
`
`edit distance calculations described in Chapter 8 of Splash2, which calculations
`
`were the basis for the Petitions in these proceedings. Rather, the cited paragraph
`
`briefly describes a paper by Gokhale and Minnich disclosing a technique for
`
`automatically synthesizing digital logic on the Splash 2 system for programs
`
`
`
`2 DirectStream also asserts, without explanation, that the Splash 2 system has some
`unidentified limitations with respect to SIMD instructions, Response, 26, but the
`cited reference (EX1011) says no such thing (in fact, it never uses the term
`“SIMD”). The claims of the 800 Patent do not mention SIMD.
`
`16
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`
`written in a language called Data-parallel Bit-serial C, or dbC. See EX2169, 37-38
`
`(pp. 14-15 of the dissertation); see EX1076¶¶15-17.3
`
`The edit distance calculations described in Chapter 8 of Splash2, however,
`
`were programmed in VHDL, EX1007, 106, not in dbC, as DirectStream’s expert
`
`confirmed on cross-examination, EX1078, 65:19-66:3; 66:12-18, as did Dr. Stone,
`
`EX1003¶150. The Gokhale and Minnich paper never mentions the edit distance
`
`application of Splash2, Chapter 8 and, as Dr. Stone explains, has nothing to do
`
`with that application described in Splash2, Chapter 8, EX1076¶¶15-17. Indeed,
`
`DirectStream does not event attempt to argue otherwise.
`
`Nor is there any question that Splash2 (EX1007) discloses forming the
`
`looping instructions within the FPGAs of the Splash 2 system. The book plainly
`
`states that “[b]oth the bidirectional and unidirectional systolic arrays have been
`
`implemented on the Splash 2 programmable logic array, with versions for DNA
`
`and protein sequences.” EX1007, 104 (emphasis added).
`
`Splash2 states that, for the bidirectional systolic array, “[e]ach processing
`
`element (PE) computes the distances along a particular diagonal of the distance
`
`matrix. A block diagram of the PE and a listing of the algorithm it executes are
`
`
`
`3 The paper is listed in Dr. Halverson’s bibliography at EX2169, 208 and in the
`record as EX1074.
`
`17
`
`
`
`IPR2018-01605, -01606, -01607
`(U.S. Patent No. 7,620,800)
`
`shown in Figures 8.6 and 8.7, respectively.” EX1007, 100. Figure 8.7 depicts the
`
`
`
`loop instructions for that array. See id., 101. Similarly, for the unidirectional
`
`array, Splash2 states “[t]he algorithm executed by each PE in the unidirectional
`
`array is listed in Figure 8.12.” Id., 104. Figure 8.12 depicts the loop instructions
`
`for the unidirectional array. See id., 105. Thus, Splash2 unequivocally discloses
`
`that it is the processing elements, o