quinn emanuel

trial lawyers | san francisco

50 California Street, 22nd Floor, San Francisco, California 94111-4788 | TEL: (415) 875-6600 FAX: (415) 875-6700

WRITER'S DIRECT DIAL NO. (415) 875-6339 WRITER'S INTERNET ADDRESS peterklivans@quinnemanuel.com

July 12, 2012 VIA ELECTRONIC FILING The Honorable Lisa R. Barton Acting Secretary U.S. International Trade Commission 500 E Street, SW Washington, DC 20436 Re: Certain Semiconductor Chips and Products Containing the Same, Inv. No. 337-TA-753

Dear Acting Secretary Barton: On behalf of Respondents, enclosed for filing is a public version of Respondents Combined Reply to Rambus and the Staffs Responses to the Commissions Notice of Review, which was filed on June 1, 2012 in the above-referenced investigation. Please do not hesitate to contact me should you have questions regarding this filing. Respectfully submitted,

/s/ Peter A. Klivans Peter A. Klivans

quinn emanuel urquhart & sullivan, llp

LOS ANGELES | NEW YORK |

865 South Figueroa Street, 10th Floor, Los Angeles, California 90017-2543 | TEL (213) 443-3000 FAX (213) 443-3100 51 Madison Avenue, 22nd Floor, New York, New York 10010-1601 | TEL (212) 849-7000 FAX (212) 849-7100 SILICON VALLEY | 555 Twin Dolphin Drive, 5th Floor, Redwood Shores, California 94065-2139 | TEL (650) 801-5000 FAX (650) 801-5100 CHICAGO | 500 W. Madison Street, Suite 2450, Chicago, Illinois 60661-2510 | TEL (312) 705-7400 FAX (312) 705-7401 WASHINGTON, DC | 1299 Pennsylvania Avenue NW, Suite 825, Washington, District of Columbia 20004-2400 | TEL (202) 538-8000 FAX (202) 538-8100 LONDON | 16 Old Bailey, London EC4M 7EG, United Kingdom | TEL +44 20 7653 2000 FAX +44 20 7653 2100 TOKYO | NBF Hibiya Building, 25F, 1-1-7, Uchisaiwai-cho, Chiyoda-ku, Tokyo 100-0011, Japan | TEL +81 3 5510 1711 FAX +81 3 5510 1712 MANNHEIM | Mollstrae 42, 68165 Mannheim, Germany | TEL +49 621 43298 6000 FAX +49 621 43298 6100 MOSCOW | Voentorg Building, 3rd Floor, 10 Vozdvizhenka Street, Moscow 125009, Russia | TEL +7 495 797 3666 FAX +7 495 797 3667 HAMBURG | An der Alster 3, 20099 Hamburg, Germany | TEL +49 40 89728 7000 FAX +49 40 89728 7100

PUBLIC VERSION

UNITED STATES INTERNATIONAL TRADE COMMISSION Washington, D.C. Before The Honorable Theodore R. Essex Administrative Law Judge _________________________________________ In the Matter of ) ) CERTAIN SEMICONDUCTOR CHIPS ) Inv. No. 337-TA-753 AND PRODUCTS CONTAINING THE SAME ) _________________________________________ )

RESPONDENTS COMBINED REPLY TO RAMBUS AND THE STAFFS RESPONSES TO THE COMMISSIONS NOTICE OF REVIEW

03990.51797/4780864.12

PUBLIC VERSION

TABLE OF CONTENTS Page I. ISSUE 1: CLAIM CONSTRUCTION (DALLY PATENTS)............................................1 A. 1(a): Output Frequency Refers to Cycles Per Second, and Cycles, in the Context of the Dally Patents, Has a 1:1 Relationship with Data Rate.....................1 1. 2. 3. 4. Fact 1 Is Incorrect: Dally Transmits One Bit Per Cycle, Not Two ..............................................................................................................2 Fact 2 Would Support Respondents and Staffs Construction................3 Rambus Construction Excludes Preferred Embodiments...........................5 Rambus Ignores Most of the Dally Specification and Relies Entirely on Two Passages that Do Not Concern Output Frequency...................................................................................................5 Rambus Mischaracterizes Poulton and Simmss Testimony.......................6

5. B.

1(b): Regardless of Whether the Commission Specifies a Data Rate for the Claimed Output Frequency, the Output Frequency Claims Are Invalid ..........7 1. 2. Widmer Alone Renders The Output Frequency Claims Obvious Under Rambus Construction ......................................................................7 The Output Frequency Claims Are Invalid Even If the Commission Declines to Require A Specific Data Rate..............................9

II.

ISSUE 2: VALIDITY .........................................................................................................9 A. B. 2(a): The ALJ Correctly Found Motivation to Combine Widmer and Ewen ..........9 Secondary Considerations......................................................................................12

III.

CERTAIN PRODUCTS DO NOT INFRINGE DALLY (ISSUES 3A, 3B, 3C)..............13 A. B. C. 3(a): Disablement of Cisco Products and SL500..................................................13 3(b): The Appropriate Standard for Infringement of the Dally Claims ................15 3(c): Rambus Failed to Prove Infringement of Method Claims 39, 40, and 42 of the 494 Patent ..............................................................................................17

IV.

THE ALJ CORRECTLY FOUND THE BARTH PATENTS INVALID IN LIGHT OF THE PRIOR ART (ISSUE 2A, 2B, 2C, AND 2D/BARTH)........................18

03990.51797/4780864.12

PUBLIC VERSION

A.

The ALJ Correctly Found the Asserted Barth Claims Obvious over Twelve Independent Combinations of Prior Art (Issues 2a) .................................18 1. 2. The ALJ Committed No Errors in Finding the Asserted Barth Claims Obvious over Each Independent Combination of Prior Art. .........18 Secondary Considerations Do Not Support a Finding of NonObviousness. ..............................................................................................26

B.

There Is No Pertinence to the Distinctions Rambus Attempts to Draw Between Allegedly Synchronous and Asynchronous Systems (Issue 2b).............26 1. 2. The Dan And Yano References Each Disclose a Synchronous System........................................................................................................27 Rambus Arguments that Persons of Ordinary Skill in the Art Would Not Seek to Combine Aspects of Synchronous and Asynchronous Systems Are Incorrect........................................................28

C. D. V. VI.

Harriman Evidences Publication of the NextBus Specification (Issue 2c)............33 Respondents Have Demonstrated that SyncLink Is Prior Art (Issue 2d)...............35

SDR PRODUCTS DO NOT INFRINGE THE BARTH PATENTS (ISSUE 3D)............36 THE ALJ CORRECTLY FOUND THE BARTH PATENTS UNENFORCEABLE DUE TO RAMBUS UNCLEAN HANDS (ISSUE 4)..................37 A. 4(a): Rambus Should be Prevented from Challenging the 661 Determinations of When It Had a Duty to Preserve Documents and Its Bad Faith .......................................................................................................................37 4(b): Respondents in this Investigation Presented Additional Factual Evidence on Prejudice Through Live Testimony and Videotaped Depositions ............................................................................................................39 1. The Record in This Investigation Reflects Substantial New Evidence on the Issue of Prejudice Adduced by Respondents at the Hearing.......................................................................................................39 Respondents Called Rambus Witnesses to Develop Additional Facts, Assess Their Credibility, and to Impeach with Prior Testimony ..................................................................................................41 The New Evidence in the Record Does Not Support Rambus Position That It Destroyed Nothing Helpful To Respondents Defenses.....................................................................................................41

B.

2.

3.

03990.51797/4780864.12

ii

PUBLIC VERSION

4. VII.

Respondents Developed New Facts In Proving Up Their New Defenses.....................................................................................................44

THE ALJ ERRED IN NOT FINDING THE BARTH PATENTS UNENFORCEABLE DUE TO RAMBUS INEQUITABLE CONDUCT (ISSUE 5) ........................................................................................................................................46 A. B. C. D. 5(a): Materiality of RX-4270C (SyncLink) ..........................................................46 5(b): Breadth of Signal and Strobe Signal Claim Elements ..........................48 5(c): Effect of PTOs Construction on Whether SyncLink is Cumulative ...........49 5(d): Scope of Unenforceability Finding ..............................................................50

VIII. IX.

ISSUE 6: DOMESTIC INDUSTRY.................................................................................52 ISSUE 7: PATENT EXHAUSTION .................................................................................55 A. B. 7(a): Licensed Samsung Memory Devices Substantially Embody the Barth Patents ..........................................................................................................55 7(b): Purchase of Samsung Memory Was in the United States.............57

X.

RAMBUS LACKS STANDING TO ASSERT THE DALLY PATENTS .......................58 A. B. 8(a): Rambus Is Not a Bona Fide Purchaser..........................................................58 8(b): Laches Cannot Defeat a Standing Challenge ................................................60

03990.51797/4780864.12

iii

PUBLIC VERSION

TABLE OF AUTHORITIES Page

CASES ACCO Brands, Inc. v. ABA Locks Mfr. Co., 501 F.3d 1307 (Fed. Cir. 2007)................................................................................... 17, 18 Advanced Display Sys., Inc. v. Kent State Univ., 212 F.3d 1272 (Fed. Cir. 2000)......................................................................................... 34 Am. Piledriving Equip. v. Geoquip, Inc., 637 F.3d Cir. 1324 (Fed. Cir. 2011) ................................................................................. 32 Bd. of Trustees of Leland Stanford Jr. Univ. v. Roche Molecular Sys., Inc., 583 F.3d 832 (Fed. Cir. 2009)............................................................................... 58, 59, 60 Boston Scientific Scimed, Inc. v. Cordis Corp., 554 F.3d 982 (Fed. Cir. 2009) ............................................................................................ 8 Certain Multimedia Display and Navigation Devices and Systems, Inv. No. 337-TA-694, Commn Op. (Aug. 8, 2011)................................................... 52, 53 Certain Personal Data And Mobile Communication Devices And Related Software, Inv. No. 337-TA-710, Final Determination (Dec. 19, 2011)............................................ 11 Chimie v. PPG Indus., 402 F.3d 1371 (Fed. Cir. 2005)........................................................................................... 5 Cook Biotech Inc. v. Acell, Inc., 460 F.3d 1365 (Fed. Cir. 2006)......................................................................................... 34 Dystar Textilfarben GmbH v. C.H. Patrick Co., 464 F.3d 1356 (Fed. Cir. 2006)......................................................................................... 21 E.I. du Point de Nemours & Co. v. Phillips Petroleum Co., 849 F.2d 1430 (Fed. Cir. 1988)......................................................................................... 32 Fantasy Sports Prop., Inc. v. Sportsline.com, Inc., 287 F.3d 1108 (Fed. Cir. 2002)......................................................................................... 17 FilmTec Corp. v. Allied-Signal, Inc., 939 F.2d 1568 (Fed. Cir. 1991)................................................................................... 59, 60 FilmTec Corp. v. Hydranautics, 982 F.2d 1546 (1992)........................................................................................................ 60 Graham v. John Deere Co., 383 U.S. 1 (1966)............................................................................................................. 19

03990.51797/4780864.12

iv

PUBLIC VERSION

In re Cyclobenzaprine Hydrochloride Extended-Release Capsule Patent Litigation, No. 2011-1399, -1409 (Fed. Cir. Apr. 16, 2012) .............................................................. 10 In re Cyclobenzaprine, 676 F.3d 1063, 2012 U.S. App. LEXIS 7571 (Fed. Cir., Apr. 16, 2012) ................... 21, 26 In re Freeman, 30 F.2d 1459 (Fed. Cir. 1994)........................................................................................... 38 In re Gurley, 27 F.3d 551 (Fed. Cir. 1994)............................................................................................. 12 In re Huai-Hung Kao, 639 F.3d 1057 (Fed. Cir. 2011)................................................................................... 13, 26 In re Natures Remedies, Ltd., 315 Fed. Appx 300 (Fed. Cir. 2009)................................................................................ 34 In re Swanson, 540 F.3d 1368 (Fed. Cir. 2008)......................................................................................... 32 In the Matter of Certain Ground Fault Circuit Interrupters, Inv. No. 337-TA-739, 2011 WL 7168646 (Dec. 20, 2011) .............................................. 37 KSR Intl v. Teleflex, 550 U.S. 398 (2007).................................................................................................... 19, 20 Kyocera Wireless Corp. v. ITC, 545 F.3d 1340 (Fed. Cir. 2008)......................................................................................... 34 Litecubes, LLC v. N. Light Prods., Inc., 523 F.3d 1353 (Fed. Cir. 2008)......................................................................................... 58 Lucent Techs. v. Gateway, Inc., 580 F.3d 1301 (Fed. Cir. 2009)......................................................................................... 18 Micron Tech., Inc. v. Rambus, 645 F.3d 1311 (Fed. Cir. 2011)......................................................................................... 40 Microsoft Corp. v. Multi-Tech. Sys., Inc., 357 F.3d 1340 (Fed. Cir. 2004)......................................................................................... 43 Moragne v. States Marine Lines, Inc., 398 U.S. 375 (1970).......................................................................................................... 39 Mothers Restaurant, Inc. v. Mamas Pizza, Inc., 723 F.2d 1566 (Fed. Cir. 1983)......................................................................................... 37 N. Am. Phillips Corp. v. Am. Vending Sales, Inc., 35 F.3d 1576 (Fed. Cir. 1994)........................................................................................... 57 Nikken, Inc. v. Robinsons-May, Inc., 51 Fed. Appx. 874 (Fed. Cir. 2002).................................................................................. 32

03990.51797/4780864.12

PUBLIC VERSION

Novo Nordisk Pharms., Inc. v. Bio-Tech. Gen. Corp., 424 F.3d 1347 (Fed. Cir. 2005) .......................................................................................... 7 Princeton Biochems., Inc. v. Beckman Coulter, Inc., 411 F.3d 1332 (Fed. Cir. 2005)......................................................................................... 10 Quanta Computer, Inc. v. LG Elecs., Inc., 553 U.S. 617 (2007).................................................................................................... 55, 56 Shell Petroleum, Inc. v. U.S., 319 F.3d 1334 (Fed. Cir. 2003)................................................................................... 37, 38 St. Clair Intellectual Prop. Consultants, Inc. v. Canon Inc., 412 Fed. Appx 270 (Fed. Cir. 2011).......................................................................... 28, 32 Standefer v. U.S., 447 U.S. 10 (1980)............................................................................................................ 38 StemCells, Inc. v. Neuralstem, Inc., Nos. 8:06-cv-01877-AW, 8:08-cv-02664-AW, 8:08-cv-01173-AW, 2012 WL 1184545 (D. Md. Apr. 6, 2012) ........................................................................................ 60 Therasense, Inc. v. Becton, Dickinson & Co., 593 F.3d 1325 (Fed. Cir. 2010) .................................................................................... 8, 46 Therasense, Inc. v. Becton, Dickinson & Comp., No. 2008-1511 (Fed. Cir. 2011).................................................................................. 46, 47 Tokai Corp. v. Easton Enters., Inc., 632 F.3d 1358 (Fed. Cir. 2011)................................................................................... 13, 26 TransCore, L.P. v. Elec. Transaction Consultants Corp., 563 F.3d 1271 (Fed. Cir. 2009)......................................................................................... 58 W. Union Co. v. MoneyGram Payment Sys., Inc., 626 F.3d 1361 (Fed. Cir. 2010) ........................................................................................ 13 Wi-Lan v. LG Elecs., Inc., 2012 U.S. Dist. LEXIS 30622 (S.D.N.Y. Mar. 7, 2012) .................................................. 18

STATUTES 19 C.F.R. 210.43.......................................................................................................................... 1 19 CFR 210.42(d) ...................................................................................................................... 38 19 U.S.C. 1337(a)(3)(c) ............................................................................................................. 52 35 U.S.C. 102............................................................................................................................... 9 35 U.S.C. 103......................................................................................................................... 9, 26

03990.51797/4780864.12

vi

PUBLIC VERSION

35 U.S.C. 261....................................................................................................................... 58, 60 35 U.S.C. 271(c) ........................................................................................................................ 37 U.C.C. 2-319(1) ......................................................................................................................... 57

OTHER AUTHORITIES RESTATEMENT (SECOND) OF JUDGMENTS 27, cmt. h..................................................................... 38

03990.51797/4780864.12

vii

PUBLIC VERSION

List of Abbreviations: Abbreviation CDX CIB CPX CPFR CPHB CRB CCN CRPFR CX ID JX NOR RDX RIB RPX RPFR RPHB RRB RCN RRPFR RRX RX SIB SRB SPFR SCN SRPFR Tr. Source Complainants demonstrative exhibit Complainants initial post-hearing brief Complainants physical exhibit Complainants petition for review Complainants pre-hearing brief Complainants reply post-hearing brief Complainants response to Commissions Notice of Review Complainants response to petitions for review Complainants exhibit ALJs Initial Determination on Violation of Section 337 and Recommended Determination on Remedy and Bond Joint Exhibit Commissions Notice of Review Respondents demonstrative exhibit Respondents corrected initial post-hearing brief Respondents physical exhibit Respondents contingent petition for review Respondents pre-hearing brief Respondents reply post-hearing brief Respondents response to Commissions Notice of Review Respondents response to petitions for review Respondents rebuttal exhibit Respondents exhibit Staffs initial post-hearing brief Staffs reply post-hearing brief Staffs contingent petition for review Staffs response to Commissions Notice of Review Staffs response to petitions for review Hearing Transcript

03990.51797/4780864.12

viii

PUBLIC VERSION

I.

ISSUE 1: CLAIM CONSTRUCTION (DALLY PATENTS) A. 1(a): Output Frequency Refers to Cycles Per Second, and Cycles, in the Context of the Dally Patents, Has a 1:1 Relationship with Data Rate

All parties agree that output frequency must be construed in terms of data rate, but disagree as to the relationship between the two. Respondents and Staff agree that output frequency in the context of the Dally patents requires a 1:1 ratio between frequency and data rate (e.g., 4 GHz equating to 4 Gbps). (See RCN at 1-2; SCN at 3.) Rambus, in contrast, would read in a limitation requiring a cycle to represent the window of time in which two signal levels, and thus two bits, are transmitted1 and thereby mandate a 1:2 ratio between frequency and data rate. (Id. (emphasis added); CCN at 1.) But Rambus position lacks merit. The Dally patents establish that a cycle cannot possibly mean the window of time in which two signals levels or bits are transmitted. Rather, as addressed at length by the Staff and Respondents and ultimately the ALJ, the intrinsic and extrinsic evidenceincluding the specification, witness testimony, and technical dictionary definitionsoverwhelmingly support construing output frequency to require a 1:1 ratio between frequency and data rate (i.e. a single bit, not two, be transmitted per cycle). (See, e.g., SCN at 3-4; SIB at 134-40; SRB at 68-72; SRPFR at 73-77; RCN at 5-9; RIB at 118-22; RRB at 55-58; RRPFR at 33-35.) To support its construction, Rambus relies on two alleged facts: (1) two signal levels are transmitted per cycle; and (2) each of those two signal levels represents one bit in the context of the Dally claims. (CCN at 1) But these are neither facts nor applicable in the context of the digital signals of the Dally patents; Rambus has constructed this artificial analysis entirely based on hypothetical signals used in other engineering contexts that are not at issue here.
1

Rambus has reversed course on this issue several times now. (Cf. CPRF at 26 (one cycle corresponds to the capability of a signal to transition high, low, and back to high) with CIB at 181 (cycle corresponds to one high signal level and one low signal level).) To the extent Rambus current window of time position differs from one of its previous positions, it should be deemed abandoned under at least 19 C.F.R. 210.43.
03990.51797/4780864.12

-1-

PUBLIC VERSION

1.

Fact 1 Is Incorrect: Dally Transmits One Bit Per Cycle, Not Two

Ignoring Dallys digital data signals, which do not repeat periodically, Rambus focuses on hypothetical single-frequency sinusoidal waveforms to demonstrate that it is possible for two signal levels [to be] transmitted per cycle, as shown for example in RDX-2189 (CCN at 2; see also, ID at 1-2, 26 (underlying Rambus construction is the assumption that the signal [] is a simple sinusoidal wave).) Electronics dictionaries define frequency as the rate at which a waveform action repeats. (Id. at 2; SIB at 138.) Pointing to its hypothetical sine waves, Rambus alleges that two signal levels, a high and a low, are constantly repeating, and thus each set of high and low levels constitute two signal levels in a single cycle of waveform action that repeats.2 (Id.) But these single-frequency sinusoidal waveforms have nothing to do with the Dally patents, which as Rambus admits, are directed to . . . improving signal integrity and data transmission. (CIB at 179 (emphasis added).) Dallys digital data signalsshown as complex, multi-frequency square wave-like signalshave properties that are completely different from Rambus simple sinusoidal waveforms, as a simple comparison of RDX-2189 and the data signals used in the patents (annotated Fig. 2A) demonstrates:

Indeed, Rambus expert Singer confirmed that, in the Dally patents as depicted in Fig. 2A above,
Rambus failed to identify how these continuous waveforms consist of two signal levels. While the waveforms depict repeating valleys and hills, Rambus makes no attempt to identify the two signal levels.
03990.51797/4780864.12
2

-2-

PUBLIC VERSION

some portions of the data signal consist of a high square wave-like signal level followed by a low signal level, whereas other portions consist of repeated bits of the same signal level: Q. Here you say figure 2A is depicting an idealized square wave pulse train, arent you, sir? A. I dont agree with that. Q. Okay. A. It says, idealized square wave-like, meaning if you focus in on a narrow section where you have one high and one low, and dont consider everything thats around it, right there it is idealized -- and imagine that it continues forever like that, then it would be an idealized square wave. But clearly no one would say figure 2A is a square wave. It clearly isnt. (Tr. at 696:4-17 (emphasis added); see also JX-120 at 7:48-49; CX-9542C at Q31.) As Singer admitted, Figure 2A clearly isnt a signal that has one high and one low repeating forever. Instead, Figure 2A shows that the only waveform action that repeats is the transmission of single bits of data, rather than pairs of high and low signal levels. As the Staff correctly noted, the plain and ordinary meaning of output frequency according to this electronics dictionary is simply the rate at which . . . data is put onto a channel by the claimed transmitter circuit. (SIB at 138-39.) Thus, each cycle in the claimed data signal transmits one bit of data,

corresponding to a 1:1 relationship between output frequency and data rate. Unlike data signals, which convey information, digital signals that merely consist of repeating 1s and 0s are classified as clock signals, are used for timing purposes, and convey no information. (See Tr. at 1479:9-12 (you are not outputting data, you are just outputting a clock signal); Tr. (Poulton) at 1775:13-15.) To define output frequency (i.e. the rate at which a waveform repeats itself) of the claimed data signals in terms of two bits per cycle makes no sense because such data will never consist of two alternating bits that repeat forever. 2. Fact 2 Would Support Respondents and Staffs Construction

Rambus also contends the claim language somehow supports its assertion that each of -3-

03990.51797/4780864.12

PUBLIC VERSION

the two signal levels represents one bit in the context of the Dally patents. (CCN at 2.) In other words, according to Rambus, the claims require that only one signal level is used to represent a single bit. (Id.) But in doing so, Rambus misinterprets the claims. In fact, Dally Figures 5C (input signal) and 5D (output signal) show that for each low or high digital value at the input, there are at least two possible signal levels on the output, not only a single level, in order to perform the claimed pre-emphasis:

To the extent Rambus contends that the claims require only one signal level on the output that corresponds to one bit in the input stream to the Dally transmitter, that is true under either construction of output frequency. In Respondents and the Staffs construction, a single signal level represents each transmitted bit, as shown in Figures 2A, 5C, and 5D. Rambus mischaracterizes the ID in alleging that the ALJ misconstrued the claims, finding that they could cover signals where multiple signal levels are required to represent one bit. (CCN at 3.) Rejecting Rambus long held and misguided assertion that the Dally patents are limited to NRZ signals,3 the ALJ concluded that other coding schemes exist within the art, such as Manchester or bi-phase coding. (ID at 27.) Applying Manchester or bi-phase encoding does not result in using multiple signal levels to represent one bit. (See RX-5431C at Q. 289.) Rambus analysis is based on comparing the inputs and outputs of the encoder. But as

Respondents expert Hassoun explained, the claims are directed to the input and output of the
3

Rambus repeatedly argued that its claim construction was based chiefly on its belief that the claims were limited to NRZ signaling. (CIB at 181; CRB at 98-99; CPFR at 26-27.) Apparently recognizing the weakness of that allegation, Rambus has dropped the NRZ argument.
03990.51797/4780864.12

-4-

PUBLIC VERSION

digital transmitter, which follows the encoder. (Id.) By the time the encoded signal reaches the digital transmitter, regardless of whether the signal is 8B/10B encoded (as in the Accused Products), Manchester encoded, or bi-phase encoded, a single bit level into the pre-emphasis circuit results in a single output level, as required by the claim language. (Id.) In this regard, the prosecution histories actually support the ALJs finding, since the Examiner had no issue with the type of encoding used by Shih, focusing instead on the unrelated issue of chip integration in discussing that reference. (See JX-122.0171.) 3. Rambus Construction Excludes Preferred Embodiments

By creating an artificial 1:2 restriction between frequency and data rate, Rambus construction conflicts with the full scope of the claims and the patent specification. Singer, for example, admitted the Dally claims cover both modulated and non-modulated signals4 (see, e.g. Tr. at 724:24-25), but conceded that Rambus output frequency construction would apply only to baseband (i.e., nonmodulated) signals (Tr. at 721:2-722:2; 722:10-14), and thus would exclude the modulated embodiments. (See, e.g., Singer Depo Tr. at 350:25-351:24 (testifying that in a modulated system, the output frequency is the center or carrier frequency, and not half the data rate).) Rambus proposed construction excludes embodiments and must be rejected. Chimie v. PPG Indus., 402 F.3d 1371, 1377 (Fed. Cir. 2005) (a construction that would not read on the preferred embodiment would rarely if ever be correct) (citations omitted.) Respondents and

Staff's construction, in contrast, cover the full claim scope, including embodiments 1 and 12 and should be adopted. (RCN at 6-7.) 4. Rambus Ignores Most of the Dally Specification and Relies Entirely on Two Passages that Do Not Concern Output Frequency

Ignoring most of the intrinsic evidence showing a direct match between frequency and

Rambus admitted this, noting Respondents argument, however, assumes the claims apply exclusively to nonmodulated signals. This is not the case. (CRB at 100)
03990.51797/4780864.12

-5-

PUBLIC VERSION

data rate, Rambus focuses on two passages that mention 4 Gbps and 2 GHz in the same vicinity. (Id. at 810:18-24; 811:25-812:5; see also CCN at 3-4; Tr. at 720:15-19 (Singer admitting the spec makes it very clear in a number of places . . . where a data rate and a single frequency are identified as aligned.) Neither supports Rambus proposed construction. As for the first, although the Dally patents state that the highest frequency of interest for a 4 Gb/s signal is 2 GHz (i.e., half of the data rate), Singer himself concedes that frequency of interest should not be interpreted as output frequency. (CCN at 4; Tr. 720:9-12, 810:7-24 (emphasis added).) Similarly, Rambus second passage refers to operating frequency, not output frequency, and neither Rambus nor Singer has established any relationship between the two. 5. Rambus Mischaracterizes Poulton and Simmss Testimony

Rambus cites to Poultons deposition testimony that [t]he maximum frequency in a four gigabit data stream is two gigahertz. (CCN at 4.) During the hearing, Poulton noted that the question eliciting that response was not very precise and that his answer applied to bandwidth, rather than frequency. (Tr. at 1726:3-6.) When asked about that testimony later, Poulton testified the only case in which youd have an output that was exactly a 1:2 ratio is if you were transmitting a [r]epeating 1010 pattern that repeats forever, which doesnt convey any information. (Tr. 1774:14 to 1775:18.) Consistent with his testimony, Figures 2 and 5 of the Dally patents show that Dallys signals are not repeating 1010 patterns. Similarly, although Rambus cites to deposition testimony from Simms related to an intermediate frequency receiver he worked on while at Texas Instruments, Rambus fails to establish any link between the particular type of signals he used and those of the Dally patents. Given that Rambus concedes output frequency depends on the type of signaling Rambus reliance on NRZ signaling in (see, e.g., CIB at 181), Rambus failure to provide this link renders Simms testimony, as well as Poultons, completely irrelevant.

03990.51797/4780864.12

-6-

PUBLIC VERSION

B.

1(b): Regardless of Whether the Commission Specifies a Data Rate for the Claimed Output Frequency, the Output Frequency Claims Are Invalid 1. Widmer Alone Renders The Output Frequency Claims Obvious Under Rambus Construction

Rambus alleges that if output frequency of at least 1 GHz is construed to correspond to a data rate of 2 Gb/s, all of Respondents validity challenges to the output-frequency claims are resolved in Rambus favor. (CCN at 5.) But Widmer in combination with Marbot, which discloses an output data rate of 3 Gbps, renders the claims obvious. (RIB at 152; RX-5431C at Q158-64.) Although Rambus contends Marbot does not disclose that signals are actually sent out of the chip at 3 Gb/s because it only states that signals are generated on-chip at that rate, this lacks merit. 5 In fact, Marbot discloses actual test measurements proving that his transmitter transmitted data at 3 Gbps, measurements that would necessarily have been generated by transmitting the signals off-chip. (RX-4410 at JNT0028329.) Moreover, even if Marbot had not provided these results, it would not matter since the law does not require actual performance of suggestions in a disclosure. Novo Nordisk Pharms., Inc. v. Bio-Tech. Gen. Corp., 424 F.3d 1347, 1355 (Fed. Cir. 2005). Rambus criticism of Widmer in view of Marbot also consists of speculative statements that Rambus characterizes as empirical evidence. (See, e.g. CCN at 7 (If Marbots technology could have been combined so easily with the Widmer Articles technology, one would have expected the Widmer Articles data rate to be much higher).) These statements should not be credited.6 Moreover, given that Rambus has never disputed that Widmer alone renders the

Rambus alleges that Marbot is not mentioned anywhere in the ID, so the ALJ apparently found all Marbot combinations waived. (CCN at 6.) However, the ALJ found in favor of Respondents construction of output frequency such that Widmer, alone or in combination with Ewen, render the claims invalid. Thus, ALJ had no need to address Widmer in combination with Marbot, which would have been redundant under Respondents claim construction. Respondents properly preserved this combination in its Petition for Review, thus, this combination has not been waived by the ALJ or Respondents. (RPFR at 64.) 6 Rambus similar statements about the Ewen combination are addressed in more detail below.
03990.51797/4780864.12

-7-

PUBLIC VERSION

claims obvious under all constructions, the ALJs finding of invalidity should be upheld and resolved entirely in favor of the Staff and Respondents. It is well accepted that a single reference, alone, can render claims obvious. (See, e.g., Boston Scientific Scimed, Inc. v. Cordis Corp., 554 F.3d 982, 990 (Fed. Cir. 2009); Therasense, Inc. v. Becton, Dickinson & Co., 593 F.3d 1325, 1337 (Fed. Cir. 2010)). Despite Respondents and the Staffs repeated assertion that Widmer alone renders the claims invalid, (see, e.g., RIB at 152; RPFR at 64; SIB at 177), Rambus has ignored this issue throughout this proceeding: a) Rambus post-hearing brief did not dispute it (CRB at 118); b) Rambus failed to challenge the issue in its petition for review (CPFR at 1, 14-17); and c) despite Respondents identifying Rambus failure to address this point in their response to Rambus petition, Rambus made no attempt to address it in its own Response (CCN at 6). Rambus, instead, simply points to the ALJs discussion of anticipation. (RRPFR at 46-47.) Thus, even if output frequency of at least 1 GHz is construed to

correspond to a data rate of 2 Gb/s, the claims are obvious over Widmer alone. Widmer discloses a transmitter circuit that can be operated either as four separate transmitters, each with a 500 Mbps output, or one transmitter that can output a single, synchronized parallel output at 2.0Gbaud, or 2 Gbps. (CIB at 150-52; see also RX-4109 at 126 (The data source provides fixed-length synchronous packets segmented into 4 parallel bytes . . . . Each link carries . . . 500 Mbaud after 8B/10B encoding.)) When operating together, the four transmitters are synchronized to transmit data as a four-bit wide parallel output signal. (See, e.g., RX-4109 at 126.) It would have been obvious to combine the multiple output signals (4) into one serial 2 Gbps output, satisfying the output frequency claims under any construction. (RIB at 152.) Combining a synchronized, parallel output into a serial output would have been easy for a person of ordinary skill using common components. For example, the preferred embodiment -8-

03990.51797/4780864.12

PUBLIC VERSION

shown in Figure 12 of the patent combines multiple parallel channels into one serial channel using a Clock Delay Line. Dally acknowledged that the Clock Delay Line used to serialize the outputs of the multiple transmitter circuits was pretty conventional and simply routine engineering to get the system to work. (JX-032C.057 at 209:6-17; JX-032C.059 at 217:8-19.) The ALJ agreed, finding that although four separate transmitters of Widmer did not anticipate the output frequency claims, Widmer rendered these claims obvious: (ID at 138, 150-51, 153 (the ALJ finds that clear and convincing evidence does show obviousness in view of the Widmer Article either alone or in combination with other art) (emphasis added).) 2. The Output Frequency Claims Are Invalid Even If the Commission Declines to Require A Specific Data Rate

Rambus alleges that [o]n the other hand, if the Commission declines to identify a data rate corresponding to the claimed output frequency . . . significant disputes will remain regarding the section 102 and 103 determinations on review. (CCN at 5; see also SCN at 6, where Staff takes a similar position.) But Rambus ignores the record establishing invalidity under Hajimiris alternative construction of output frequency, which does not specify a data rate. As Respondents discussed in detail, Hajimiri opined that output frequency if not

indefinite refers to the frequency content of the output signal. (RCN at 10.) Rambus does not contest that the SL500 and the Widmer article anticipate the output frequency claims under Hajimiris construction. Thus, to the extent that the Commission finds in favor of Hajimiris alternative construction, the output frequency claims are also invalid. II. ISSUE 2: VALIDITY A.
7

2(a): The ALJ Correctly Found Motivation to Combine Widmer and Ewen7

Respondents note that the Commissions decision regarding the combination of Widmer and Ewen, which affects the claims requiring differential signaling and output frequency, should not disturb the ALJs finding that these claim are obviousness. The ALJ relied on grounds independent of the Widmer and Ewen combination to find obviousness that Rambus has not requested for review, and thus waived. (See RRPFR at 46-47.)
03990.51797/4780864.12

-9-

PUBLIC VERSION

Rambus relies almost entirely on conjecture and trivial differences between the two references to argue no motivation to combine.8 For example, Widmer states that the disclosed serializer/deserializer (SerDes) design adopts techniques used previously. (RX-4109 at 126.) Rambus divines an entire paragraph of speculative conclusions for this simple statement. Relying on Singer, Rambus asserts that [t]he authors of the Widmer Article already claim to have incorporated the relevant features of [Ewen] into their design, so one of ordinary skill in the art would not have considered combining other features from that paper. (CX-10764C at Q122 (emphasis added).) Rambus continues, alleging that if there were additional techniques to be imported from Ewen, the Widmer Articles team would have already known about them and incorporated them. (CCN at 9 (emphasis added).) Rambus then speculates that the Widmers team had already used their higher level of skill to incorporate what they could. (Id.

(emphasis added).) Rambus essentially reads the phrase adopts techniques used previously to mean already adopts any and all techniques used previously and any techniques not already adopted should not be adopted. Clearly, this contradicts the plain language of Widmer. Rambus speculates, without a scintilla of evidence, that the authors of the Widmer Article desired to achieve a 1062 MBaud output in the Widmer chip, attempted to do so, and failed. But Rambus offers no evidence for these assertionsand it cannot, having failed to depose any of the authorsand ignores the myriad possible reasons (e.g., target cost, likely use scenarios) as to why the authors of the Widmer Article chose to design a quad transceiver operating at 500 Mbps
Rambus cites Princeton Biochems., Inc. v. Beckman Coulter, Inc., 411 F.3d 1332, 1337 (Fed. Cir. 2005) to support is position that if increased speed alone were sufficient to justify combining features of the prior art, the patents of most industries would be in jeopardy. CCN at 9. But Princeton fails to address speed at all; the Federal Circuit found the claims obvious because, as here, the motivation to combine these references was found in the knowledge of those skilled in the art. Id. at 1338. Rambus reliance on In re Cyclobenzaprine Hydrochloride Extended-Release Capsule Patent Litigation, No. 20111399, -1409 (Fed. Cir. Apr. 16, 2012) is equally misplaced. The Federal Circuit based its finding on the fact that there is no evidence that skilled artisans would have known how to achieve the solution. Id. at 19. Here, in contrast, as Hassoun testified, it would have been quite easy for a person of ordinary skill to combine the teachings of Ewen and Widmer. (Tr. at 1501:8-15.)
03990.51797/4780864.12
8

-10-

PUBLIC VERSION

data rate to be used in the particular product discussed in the Widmer Article. (RRB at 77.) Although Rambus assumes designers would always use those techniques that result in the fastest products, without consideration of other design factors, and the Accused Products, which include parts with a variety of different speeds, refutes this. (See, e.g. CX-9542C at Q. 78.) Rambus asserts that it is a fact that Widmers team, which included all the authors of Ewen, still could not achieve the speeds described in Ewen. (Id. at 9.) But Rambus

acknowledges on the very next page that that the authors of Ewen, who were part of Widmers team, achieved Ewens data rate of 1.062 Gbps one year before Widmer was published. (Id. at 10.) Compounding its error, Rambus asserts that this fact compels the conclusion that it would not be obvious to combine the two references to achieve a higher speed. Rambus also relies on a host of trivial differences that it alleges prevents a person of ordinary skill from combining Widmer and Ewen, such as [s]erial-link technology is highly sensitive, Widmer is single-ended while Ewen is differential, and Widmers design was fabricated in 0.5m CMOS while Ewen was fabricated in 0.8m CMOS. (CCN at 11.) The Commission has rejected arguments proposing that the prior art be like two puzzle pieces that must fit together perfectly. Certain Personal Data And Mobile Communication Devices And Related Software, Inv. No. 337-TA-710, Final Determination (December 19, 2011). Moreover, all of these differences would have been quite easy to overcome for a person of ordinary skill in the art. (Tr. at 1501:8-15.) For example, one of ordinary skill would have recognized the increased speed of Ewen was at least partially a result of using differential outputs, instead of the single-ended outputs of Widmer. (See, e.g., CX-6683C.003 (Poulton stating that [t]he system Im currently working on is differential. Single-ended is just too difficult for speeds in excess of 500 Mbit/sec or so..) One of ordinary skill would have been able to perform the trivial task of switching between -11-

03990.51797/4780864.12

PUBLIC VERSION

differential and single-ended outputs. (RIB at 149; RX-5431C at Q. 194.) Tellingly, the Dally patents are silent on these issues and fail to teach how to address any alleged sensitivity, semiconductor process dependency, or means of switching between single-ended and differential signals, which is significant in light of the claims requiring a semiconductor chip and encompassing single-ended signaling. (RRB at 79.) Rambus also alleges that Widmer teaches away from differential signaling. (CCN at 11.) But [a] known or obvious composition does not become patentable simply because it has been described as somewhat inferior to some other product for the same use. In re Gurley, 27 F.3d 551, 553 (Fed. Cir. 1994). Widmer teaches that a single-ended output can be used to reduce wire count. One of ordinary skill would have been well aware of this and other benefits and

drawbacks of single-ended versus differential signaling. (RIB at 149-50; see, also, RX-5431C at Q. 194; CX-6683C.003.) To use one over the other would have been nothing more than a design choice. A transmitter using differential signaling does not become patentable simply because Widmer chose to use a single-ended output for a particular system. Indeed, although Dally

himself, during prosecution, suggested differential signaling offers better noise immunity (RX2239.0198), many of the asserted claims cover single-ended signaling. The Dally prosecution history can hardly be interpreted to teach away from techniques it claims. B. Secondary Considerations

Rambus petition did not address the issue of secondary indicia of nonobviousness for the Dally patents and, accordingly, that issue has been waived. To the extent the Commission is inclined to address this issue on its own initiative, Respondents note that the ALJ properly considered secondary considerations in determining obviousness, but ultimately found that the evidence was outweighed by the strong showing of obviousness. (ID at 228-31.) [W]eak secondary considerations generally do not overcome a strong prima facie case of obviousness.

03990.51797/4780864.12

-12-

PUBLIC VERSION

W. Union Co. v. MoneyGram Payment Sys., Inc., 626 F.3d 1361, 1373 (Fed. Cir. 2010). Indeed, even if Rambus could establish the required nexuswhich it cannota highly successful product alone would not overcome the strong showing of obviousness. Id. (citations omitted). As Respondents have shown, Rambus has failed to establish any secondary indicia of non-obviousness. (RCN at 22-24.) Rambus relies heavily on the unsubstantiated testimony from Poulton, much of which Poulton himself later clarified based on context. (Id.) Where the inventions represented no more than the predictable use of prior art elements according to their established functions, the secondary considerations advanced by Rambus are inadequate to establish nonobviousness. Id.; see also In re Huai-Hung Kao, 639 F.3d 1057, 1068 (Fed. Cir. 2011); Tokai Corp. v. Easton Enters., Inc., 632 F.3d 1358, 1369 (Fed. Cir. 2011); see also SCN at 20-26 (rejecting Rambus evidence of secondary considerations of non-obviousness). III. CERTAIN PRODUCTS DO NOT INFRINGE DALLY (ISSUES 3A, 3B, 3C) A. 3(a): Disablement of Cisco Products and SL500 Rambus arguments regarding the functionality of the SeriaLink SL500 technology and its embodiment in one particular product (the SL500 Fiber Channel Transceiver) are largely false and generally misleading. First, Rambus asserts that the SL500 is incapable of providing emphasis. (CCN at 42.) Specifically, Rambus argues that inputs to the SL500FCT product were fixed such that there was no emphasis and that there was nothing a user could do to adjust or modify the inputs to enable emphasis. (Id.) This is false and was properly rejected by ALJ Essex. Rambus only support for these assertions is a July 1996 document that describes one variation of an SL500 chip. (See RX-4171; Tr. (Hajimiri) at 1615:10-1616:1.) But this documentoffers no support for Rambus assertion that the emphasis inputs were set in a manner that would prevent a user from enabling emphasis in the chip. (See generally RX-4171.) Moreover, as Respondents expert

03990.51797/4780864.12

-13-

PUBLIC VERSION

Hajimiri pointed out, the single document relied upon by Rambus merely describes one instantiation of the SL500 transmitter. (See RX-4171; Tr. (Hajimiri) at 1615:10-1616:1.)

Abundant evidence, including actual test data and contemporaneous business records, showed that LSI offered for sale and sold to Seagate (the SL500 chip) and Cabletron (the Shemp chip) products that operated using emphasis. (See, e.g., RX-4130C; RX-4116C; RX-4156C; RX4167C; RX-4149C; RX-4150; RX-4175C; RX-4173C; RX-4128; RX-4142C; RX-4137C; RX4136; RX-4138.) The emphasis functionality in the SL500 was programmable and could be set to one of multiple values, including a low value. (RX-4130C.0047.) This does not mean that its emphasis functionality was inoperable. Rambus blanket statement that emphasis

functionality was simply not present in the SL500 semiconductor chip is incorrect as to the SL500FCT and in direct conflict with the plethora of evidence in the record demonstrating that LSI Logic developed, offered for sale, and sold the SL500 with emphasis. (See, e.g., RX4173C.0048; RX-4173C.0065.) Second, Rambus argues that the emphasis functionality did not operate properly. Again, Rambus argument misstates the record and properly was rejected by ALJ Essex. Rambus expert, Singer, testified that a simulation showing an output waveform generated by the SL500 transmitter illustrates its nonfunctionality, apparently due to the presence of voltage overshoots. (CCN at 42-43; see also CX-10764C.0129-30.) In fact, the simulation waveform proves exactly the opposite: the transmitters emphasis circuitry was operating properly and entirely as desired and expected. (RX-5430C at Qs. 491-496.) Hajimiri testified that the pre-emphasis circuit was designed so as to exhibit voltage overshoots in what is known as an under-damped design. (Id.) The simulation shows that the transmitter is doing exactly what it supposed to do by emphasizing the first bit after transition and de-emphasizing the subsequent bits of the same value. (Id.) Further, contrary to Rambus claim, the simulated waveform never crosses the reference -14-

03990.51797/4780864.12

PUBLIC VERSION

threshold at any point during the simulation. (Id.; see also RDX-350.) Accordingly, as the ALJ correctly concluded (ID at 190), the emphasis functionality in the SL500 was entirely operational, a conclusion that was confirmed by a Seagate within months of the collection of the abovereferenced simulation data. (See RX-4173C.0159-160

) The Cisco products without functioning transmitters, however, do not practice the Dally claims. As explained previously, certain Cisco products contain components from the Supplier Respondents that have transmitters accused of infringing the Dally patents. (See RCN at 27-28.) For both Rambus and Staff, that Ciscos products include the accused transmitters and their preemphasis circuitry is enough to prove infringement because those transmitters remain capable of infringing use. (See CCN at 43-44; SCN at 43.) In Ciscos products, however, the accused transmitters are not used, nor can they be used, because those transmitters are not connected to anything. Even Rambus expert conceded that products in which the accused transmitters are not connected do not infringe the Dally patents because the disconnected transmitters are not operable to send any signals. (See Tr. (Singer) at 770:21-771:2 (In that case, if not pinned out meant that there werent any wires connecting that circuitry, and it was disabled in that fashion, yes, I would agree it is disabled and bypassed and it cannot infringe.).) Ciscos products in which the accused transmitters are not connected to anything do not infringe.9 B. 3(b): The Appropriate Standard for Infringement of the Dally Claims

Rambus bases its infringement analysis on an erroneous interpretation of the asserted

An analogy in Rambus brief illustrates the point. Rambus states that Ciscos argument is akin to stating that its products are not operable to send or receive signals because they are not connected to a power source when sold. (CCN at 44.) On the contrary, Ciscos products could be plugged in, fully configured and installed, and the accused transmitters still would not be capable of sending any signals, because they are not connected to anything.
03990.51797/4780864.12

-15-

PUBLIC VERSION

Dally claims. Rambus characterization of the asserted claims as requiring a transmitter circuit that is operable to send an output signal having certain features is misleading. (CCN at 44). Claim 1 of the 857 patent claims the transmitter circuit being operable to send an output signal so that: (i) an output bit signal of the output signal representing a particular bit value has one signal level when the bit value represented by a predetermined preceding output bit signal; and (ii) the output bit signal representing the particular bit value has another signal level when the bit value represented by the predetermined preceding output signal level. (See JX-120 at 8:56-64.) In making its infringement allegations, Rambus fixates on the claim term operable to, but ignores the substance of the claimthe comparison of a current bit to a preceding bit. There is no suggestion of using succeeding (future) bits of the input data to perform preemphasis, either in the claims or the specification of the Dally Patents. (See generally JX-120.) Thus, infringement of the asserted claims of the Dally patents is dependent upon a comparison ofa current bit to a preceding bit to determine the level of a transmitters output signal. In contrast to the asserted Dally claims, products that incorporate a use a transmitter that uses current, preceding, and succeeding bits to determine the level of a transmitters output signal. (RCN at 29.) Rambus has conceded that products with functional are outside the scope of the asserted Dally claims. (See Tr. (Singer) at 752:524.) Therefore, Rambus bases its infringement allegations on a scenario in which one of the is set to zero because according

coefficients of a

to Rambus, a 3-tap filter [can] operate in the same manner as a 2-tap filter. (CIB at 197; see also Tr. (Singer) at 752:25-753:1; CPHB at 291-92; CRB at 105.) However, products that incorporate a never can operate as a 2-tap filter. The are

always operational, even when they are assigned a 30, 137-60; RX-6243 at Qs. 26-27.) -16-

(See RX-6244 at Qs. 112-

03990.51797/4780864.12

PUBLIC VERSION

By Singers own admission, even if the the circuitry associated with the taps remains unchanged. (Tr. (Singer) at 752:12-13.) As such, the circuitry for that tap never disabled and renders a outside

the scope of the claims of the Dally Patents. (RX-6244C at Q123-24, 127-28; Tr. (Singer) at 750:4-16; RX-6243C at Qs. 35-39; Tr. (Tracy) at 1405:18-24, 1406:21-1407:1.) The ALJ also mistakenly found infringement of products stating that

when the coefficient corresponding to a succeeding bit is set to zero, that tap has no impact on the output signal level. (ID at 75.) However, this finding fails to take into account the language of the claims, as well as the nature of the accused product Fantasy Sports Prop., Inc. v. Sportsline.com, Inc., 287 F.3d 1108, 1118 (Fed. Cir. 2002). Even if consideration of a succeeding bit had "no impact on the output signal level," it is irrelevant to the infringement analysis. The claims are not directed to an "impact" on the signal level. succeeding bit is used in the calculation of the output signal level takes filters outside the scope of asserted Dally claims. Thus, according to the Federal Circuits rationale in Fantasy Sports and ACCO Brands, Inc. v. ABA Locks Mfr. Co.10 (see RCN at 28-29), products that incorporate 3 and 4-tap filters should be found to be noninfringing. C. 3(c): Rambus Failed to Prove Infringement of Method Claims 39, 40, and 42 of the 494 Patent11 The fact that a

10 11

501 F.3d 1307 (Fed. Cir. 2007). Rambus also failed to address in its opening brief STs showing before the ALJ and in its post-hearing brief that there is insufficient proof that ST induced infringement of the Dally method claims. (See CCN at 47 n.11.) Contrary to Rambus statement in its brief, ST did raise and, therefore, did not waive the issue of lack of proof of induced infringement in its post-hearing brief. (See RCN at 30-31 (citing RPHB at 134).) As for Staffs argument that the evidence is sufficient to prove induced infringement, it relies on Lucent Techs., Inc. v. Gateway, Inc., 580 F. 3d 1301, 1317 (Fed. Cir. 2009), but the evidence in that case
03990.51797/4780864.12

-17-

PUBLIC VERSION

For similar reasons discussed above, the Commission should find that products incorporating 3 or 4-tap filters do not infringe claims 39, 40, and 42 of the 494 patent. These claims disclose a method of using [a] transmitter to drive the series of output bit signals to have magnitude represented by whether the first bit represents a common logic level relative to an immediately previous bit (JX-120 at col. 12:14-16.) With respect to method claims that if one or more steps of a claimed method are not performed as recited, the claim is not infringed. See Lucent Techs. v. Gateway, Inc., 580 F.3d 1301, 1317 (Fed. Cir. 2009). The language of claims 39, 40, and 42 of the 494 patent clearly requires comparison of only two bitscurrent and preceding. As previously discussed, Products that incorporate 3 and 4 tap filters compare the current, preceding, and succeeding bits in order to determine the magnitude of the output bit signal. The use of a third tap regardless of whether the tap is set to zero or a non-zero coefficientrenders these products outside the scope of these claims. Thus, the steps of claims 39, 40, and 42 of the 494 patent are not performed as claimed and the Commission should find that these products do not infringe these claims.12 IV. THE ALJ CORRECTLY FOUND THE BARTH PATENTS INVALID IN LIGHT OF THE PRIOR ART (ISSUE 2A, 2B, 2C, AND 2D/BARTH) A. The ALJ Correctly Found the Asserted Barth Claims Obvious over Twelve Independent Combinations of Prior Art (Issues 2a) 1. The ALJ Committed No Errors in Finding the Asserted Barth Claims Obvious over Each Independent Combination of Prior Art.

In this Investigation, the ALJ identified twelve separate and independent combinations of prior art, each of which standing alone renders every asserted claim of the Barth patents invalid

was barely sufficient. Id. at 1318. The evidence here does not even meet that meager standard. (See RPHB at 134.) 12 Rambus assertion of indirect infringement against end-users is based almost exclusively on conjecture by its own expert. (CCN at 47; see also CX-9542C.00498-500.) However, this is insufficient evidence to prove direct or indirect infringement of a method claim. See Wi-Lan v. LG Elecs., Inc., 2012 U.S. Dist. LEXIS 30622, *17-18 (S.D.N.Y. Mar. 7, 2012); ACCO Brands, 501 F.3d at 1313.)
03990.51797/4780864.12

-18-

PUBLIC VERSION

as obvious. Nonetheless, Rambus contends that the ALJ got it wrongnot once, not twice, but twelve different times. In reality, the ALJ got it right, and the Commission should affirm the ALJs holding that all of the asserted Barth claims are invalid as obvious over the prior art. In finding the Barth claims invalid, the ALJ properly analyzed in considerable detail each of the factors required by the Supreme Court in determining obviousness, including the scope and content of each of the prior art references. See Graham v. John Deere Co., 383 U.S. 1, 17 (1966); ID at 90-123, 210-23.13 Moreover, just because the ALJ incorporated by reference into his obviousness section many sections from his previous 34-page anticipation discussion of the scope and content of the various Barth prior art references (see ID at 210-23 incorporating sections from ID at 90-123), does not mean that ALJ legally erred by grouping references without discussing or distinguishing what they disclose as Rambus argues. (CCN at 12.) In reality, the ALJ not only evaluated the scope and content of the prior art, but commendably consolidated and focused his Barth obviousness analysis on the key issues in dispute between the parties in the Investigation. (ID at 210-23.) Similarly, Rambus criticism that the ALJ somehow legally erred by not considering the claimed inventions as a whole has no merit. (CCN at 12.) In the ID, the ALJ consistently evaluated whether the subject matter [of the Barth claims] as a whole would have been obvious as required by 103, repeatedly finding that the patented inventions as a whole consist of nothing more than combinations of well-known methods performing established function[s] that do[] no more than yield predictable results, and therefore such combinations would have been obvious to one of ordinary skill in the art. (ID at 213-14; 215; 216; 218; 22021; and 223 (citing KSR Intl v. Teleflex, 550 U.S. 398, 417 (2007).) Any grouping of claim
Indeed, Rambus allegation that [i]n the lengthy ID, there is virtually no discussion of SCI and Release 4 (CCN at 12) is demonstrably incorrect. (See, e.g., ID at 213 (analyzing Figure 6 of SCI); ID at 214 (analyzing Table 3.11.5-1 of JEDEC Release 4).)
03990.51797/4780864.12
13

-19-

PUBLIC VERSION

features performed by the ALJ simply reflected the obvious overlap between like elements used across the various asserted claims, and also illustrated the ubiquity of those claimed features in the cited prior art consistent with the Supreme Courts guidance in KSR. (See id.) Moreover, the ALJ did not premise his finding that the claimed features were wellknown solely on Jacobs testimony as Rambus suggests (CCN at 12-13), but also based his conclusions on the express content of the references cited by Jacob in his witness statement. The content of those references plainly discloses that each and every element of the asserted Barth claims was well known in the prior art. (See ID at 90-123, 210-23). As such, the fact that certain prior art references evaluated by the ALJ in this Investigation were not considered or addressed by Jacob in his textbook is irrelevant. The references were before the ALJ. And the references clearly and convincingly demonstrate that the asserted claims consist of nothing more than combinations of well-known methods performing established function[s] that do[] no more than yield predictable results, and therefore such combinations would have been obvious to one of ordinary skill in the art. (ID at 213-14, 215, 216, 218, 220-21, and 223.) Rambus also mistakenly accuses the ALJ of improperly relying on hindsight because the ALJ purportedly cited similar improved speed and faster data access reasons for combining references in invalidating multiple patent families. (CCN at 13-14.) Yet what the ALJ may have decided on other patents is irrelevant, as the ALJ documented that the improved efficiency and faster data access reasons for combining the invalidating Barth prior art are expressly disclosed in many of the references themselves. (ID at 218-19; RX-5429C (Jacob Direct) at Q. 489.) And even if it was somehow relevant, the fact that similar reasons exist for combining references with respect to the cited patent families is not surprising, as universal motivations such as improved speed and faster data access are implicit in this technology space. See Dystar Textilfarben GmbH v. C.H. Patrick Co., 464 F.3d 1356, 1368 (Fed. Cir. -20-

03990.51797/4780864.12

PUBLIC VERSION

2006); see also ID at 213, 214-15, 216, 218, and 222 (each citing RX-5357 at 28). Moreover, as discussed, the ALJ cited more than improved speed and faster data access as reasons for combining references, finding also that the asserted Barth claims merely involve well-known methods from the prior art performing established function[s] that do[] no more than yield predictable results, and therefore such combinations would have been obvious to one of ordinary skill in the art. (ID at 213-14, 215, 216, 218, 220-21, and 223.) Similarly, Rambus misapplies a recent Federal Circuit panel decision concerning pharmaceutical formulations to the ALJs findings in this Investigation regarding electronic memory controllers. (See CCN at 14 (citing In re Cyclobenzaprine, 676 F.3d 1063, 2012 U.S. App. LEXIS 7571 (Fed. Cir., April 16, 2012)).) In Cyclobenzaprine, the panel observed that an FDA document that merely provides guidance to sponsors/applicants on what they should do if they plan to pursue bioequivalence provides little support for obviousness where there is no evidence that a skilled artisan would have targeted bioequivalence in the first instance. Id. at *26-27. In contrast, the prior art relied upon by the ALJ expressly evidences that skilled artisans not only were targeting improved speed and increased data access, but had developed a set of well-known features (e.g., DDR, strobe signals, etc.) that achieved this aim. (See ID at 90-123, 213-223.) As such, the ALJs twelve independent combinations of prior art features performing established function[s] that do[] no more than yield predictable results are far different from the hypothetical combination in Cyclobenzaprine where there is no evidence that skilled artisans would have known how to achieve [a pharmaceutical extended-release formulation]. See Cyclobenzaprine at *27. (a) The ALJ Correctly Found that Combinations 1-614 Each Render the Asserted Barth Claims Invalid.

14

References to Combinations refer to the twelve independent combinations of Barth prior art references listed in the summary table on pages 210-11 of the ID.
03990.51797/4780864.12

-21-

PUBLIC VERSION

Rambus premises most of its arguments regarding Combinations 1-6 on an alleged distinction between synchronous and asynchronous references. (CCN at 14-17.) Yet, all of these arguments necessarily fail because both Yano and Dan are synchronous memory systems. (RCN at 48-52; see also RX-5429C (Jacob Direct at Qs. 491-501.) Therefore, Rambus various arguments against combining either Yano or Dan with the disclosure of Farmwald 037, SCI, and/or Release 4 are inapplicable. (CCN at 14-16.) Moreover, even if Yano and Dan were asynchronouswhich they are notthis distinction would not matter because persons of ordinary skill look to both synchronous and asynchronous systems to solve problems in the art. (RX-5429C at Q496.) Indeed, Rambus primary evidence to the contrary is the unsupported testimony of its expert Przybylski (CCN at 14-17 (repeatedly citing CX-10765C))testimony the ALJ found so unreliable overall that his testimony in this matter could not assist him. (ID at 309.) Therefore, the ALJ correctly determined that [t]he evidence shows that it would indeed have been obvious to combine aspects of asynchronous and synchronous art so as to achieve the inventions claimed in the Barth I patents. (ID at 212 (citing RX-5429C at Q105, 129, and 49296).) The Commission also should disregard Rambus allegation that somehow the asserted Barth claims treat as important decoupling and supposedly Dan and Yano do not. (CCN at 14-15.) In reality, Rambus position on whether the asserted claims require decoupling varies wildly depending on whether Rambus is arguing infringement or invalidity. Specifically,

Rambus currently argues with respect to invalidity that the asserted claims of the Barth patents treat as important decoupling the timing between a command and data transfer through use of a strobe signal. (CCN at 15.) Yet this position directly contradicts Przybylskis infringement testimony at trial in which he swore that the claims dont recite the decoupling element that you are implying. (Tr. 869:21-870:14 (emphasis added).) In other words, Rambus argues that -22-

03990.51797/4780864.12

PUBLIC VERSION

decoupling is not required for infringement,15 but somehow is required for invalidity. Like the ALJ, the Commission should disregard such blatantly inconsistent arguments from Rambus and its expert. (ID at 305-09; see also ID at 121 (criticizing Rambus inconsistent arguments and holding that Rambus cannot construe strobe signal in two different ways to prove infringement and avoid invalidity.) In addition, Rambus incorrectly argues that SCI also fails because it is not prior art. (CCN at 16.) The evidence, however, refutes Rambus argument. In particular, SCI is dated September 1989 and was presented at the Eurobus conference in London, England. (RX4278C.0001; RX-5429C at Q456.) SCI also states the Scalable Coherent Interface (IEEE P1596) will establish an interface standard for very high performance multiprocessors and invites if you would like to participate in this work, or if you would like more detailed information, please contact one of the authors or the chairman of the project and provides contact information. (RX-4278C.0001, .0008.) On the basis of such evidence, the ALJ properly found that SCI is prior art to the Barth patents. Rambus also erroneously argues that one of ordinary skill would not combine SCI with Yano or Dan because SCI purportedly does not describe sending data to memory devices. (CCN at 16.) Yet this statement, in addition to mischaracterizing the SCI reference, is legally unsound because it attacks the combination of the complete system disclosed in SCI with Yano and/or Dan, instead of properly assessing whether the singular element disclosed in SCI the DDR claim elementwould have been obvious to combine with Dan and/or Yano. 16 (RX5429C (Jacob Direct) at Qs. 499-501.) Therefore, the ALJ did not err in finding the asserted
15

The reason Rambus must argue that decoupling is not required by the asserted claims in the context of infringement is that the accused products are not decoupled and, therefore, would not infringe if this was a requirement. (RIB at 31-32; RRB at 10-11.) 16 Indeed, the BPAI chided Rambus for employing similar improper legal arguments during the reexamination of the Barth 405 patent. (RX-5357.0013-14.)
03990.51797/4780864.12

-23-

PUBLIC VERSION

Barth claims invalid over combinations of prior art involving SCI. Similar to its argument with respect to SCI, Rambus also argues that one of ordinary skill in the art would not have incorporated the alleged strobe signals from the slower, asynchronous systems of Yano and Dan into the already faster systems of Farmwald. (CCN at 17.) Yet, besides being premised on the false characterization of Yano and Dan as asynchronous, this argument again improperly targets the motivation to combine two systems instead of properly assessing whether the singular element disclosed in either Yano and Danthe claimed strobe signalwould have been obvious to combine with Farmwald 037. (RX-5429C (Jacob Direct) at Qs. 499-501.) Therefore, the ALJ did not err in finding the asserted Barth claims invalid over combinations of Farmwald 037 with the strobe signal from either Dan or Yano. (b) The ALJ Correctly Found that Combinations 7-12 Each Render the Asserted Barth Claims Invalid.

Rambus criticisms of the ALJs findings with respect to Combinations 7-12 also are unfounded. First, the ALJ did not fail[] to address why it would have been obvious to

implement the alleged DDR features in NextBus or Harriman as Rambus argues. (CPFR at 54.) Although not acknowledged by Rambus, the ALJ actually found that DDR already is disclosed in both NextBus and Harriman, thus mooting any need whatsoever to implement the DDR features in NextBus and Harriman. (ID at 220; CPFR at 54.) Furthermore, it would have been obvious to one of ordinary skill in the art to modify the DSTB* signal of the NextBus Specification and Harriman to sample on both sides of the clock because dual edge clocking was well-known in the art, as taught in Farmwald 037 and SCI. (RX-5429C (Jacob Direct) at Qs. 469, 507.) Again, the only evidence Rambus cites to contest this finding are naked assertions from its expert Przybylski as to what one of ordinary skill in the art would have or would not have done (see CPFR at 54)assertions the ALJ expressly rejected. (ID at 306.)

03990.51797/4780864.12

-24-

PUBLIC VERSION

Rambus also wrongly alleges that certain claim elements (e.g., precharge commands) are irrelevant to the backplane bus of NextBus and Harriman and that the ALJs construction of memory device yields a non-sensical result. In reality, a memory controller can easily send precharge or other commands to a memory board, and nothing in the asserted claims requires the precharge command or other signals to be generated on the memory board disclosed in the NextBus Specification as Rambus alleges. (RX-5429C (Jacob Direct) at Qs. 509-511, 528-31.) Moreover, Rambus argument that the system of Farmwald 037 cannot be combined with the DSTB* strobe signal from NextBus or Harriman because Farmwald 037 purportedly runs twenty times faster is irrelevant. (CCN at 19.) The ALJ did not find that one of ordinary skill would include the clock signal from the NextBus system in the memory system of Farmwald 037, but rather the strobe signal DSTB* which can operate at whatever speed the system clock operates. (ID at 222-23.) Thus, in the combined system, the DSTB* signal (from the NextBus Specification or Harriman) would be used to sample data instead of the clock (as in Farmwald 037) in order to minimize clock skewthe precise reason NextBus and Harriman each use DSTB* to sample data instead of the clock and the same reason DQS is used to sample data in the Accused Products. (RX-5429C (Jacob Direct) at Qs. 296, 526.) For this reason alone the ALJ correctly found it would have been obvious to one of ordinary skill in the art to combine those elements of NextBus Spec or Harriman with the disclosure of Farmwald 037, thereby invalidating all of the asserted Barth claims. (ID at 222.) Finally, Rambus ignores efficiency in criticizing the ALJs finding that combinations with Farmwald 037 would improve the efficiency and speed of data access operations by indicating to the target memory that data is on the bus. (Id.) Specifically, Rambus argument that Farmwald 037 already solved the problem with a delay value ignores the fact that a delay value is inefficient because it is subject to clock skewthe same skew that is minimized -25-

03990.51797/4780864.12

PUBLIC VERSION

by using a strobe signal instead of a delay value. (See RX-5429C (Jacob Direct) at Q. 526.) Therefore, the ALJ properly found that one of ordinary skill in the art would seek to combine the strobe signal of either NextBus or Harriman with the operation method of Farmwald 037. (See ID at 222.) 2. Secondary Considerations Do Not Support a Finding of NonObviousness.

In reference to the asserted Barth claims, the ALJ properly considered secondary considerations in determining obviousness, but ultimately found that the evidence was outweighed by the strong showing of obviousness in this instance. (ID at 226.) And Rambus citations to unsubstantiated testimony from Przybylski does not compel a different result (CCN at 27-28 (citing CX-10765C ).) This is because Rambus does not and indeed cannot establish the required nexus between any supposed evidence of secondary considerations (i.e., commercial success and licensing) and anything novel in the asserted Barth claims. In particular, every element of the asserted claims was well-known in the prior art, including the strobe signal. See, e.g., In re Huai-Hung Kao, 639 F.3d 1057, 1068 (Fed. Cir. 2011); Tokai Corp. v. Easton Enters., Inc., 632 F.3d 1358, 1369 (Fed. Cir. 2011). Therefore, this lack of reliable evidence regarding objective or secondary considerations of non-obviousness, when considered contemporaneously with the evidence supporting the twelve independent combinations of prior art found invalidating by the ALJ, clearly and convincingly demonstrates that the asserted Barth claims are invalid under 35 U.S.C. 103. Cf. Cyclobenzaprine, 2012 U.S. App. LEXIS at *43 (holding that all relevant evidence must be considered before judgment on an obviousness challenge.) B. There Is No Pertinence to the Distinctions Rambus Attempts to Draw Between Allegedly Synchronous and Asynchronous Systems (Issue 2b)

For the reasons set forth in Respondents Response to the Commissions Notice of

03990.51797/4780864.12

-26-

PUBLIC VERSION

Review, there is no pertinence to the distinctions Rambus attempts to draw between allegedly synchronous and asynchronous systems. (RCN at 48-55.) Specifically, the only primary

references the ALJ relied on in rendering his obviousness findings (i.e., Dan and Yano) are synchronous references. (Id. at 48-51.) Therefore, Rambus alleged distinctions between

synchronous and asynchronous systems are irrelevant to this Investigation. In addition, even assuming this supposed distinction is relevant to the prior art combinations involving Dan and Yano, the overwhelming evidence demonstrates that a person of ordinary skill in the art would have been motivated to apply the teachings of both asynchronous and synchronous systems in developing memory systems. 1. The Dan And Yano References Each Disclose a Synchronous System

Rambus allegations regarding the differences between asynchronous and synchronous are only relevant to the ALJs obviousness findings if the Dan and Yano references are in fact asynchronous references, which they are not. Rambus, however, all but ignores this threshold issue in its response to the Commissions Notice of Review. Indeed, in its one paragraph response to this issue, Rambus alleges that Dan and Yano are asynchronous references because these references do not show a clock signal from the memory controller to the memory device. (CCN at 33.) Rambus position is incorrect for at least two reasons. First, whether a memory device receives a clock signal directly from the memory controller is not determinative of whether a memory system is synchronous or asynchronous. As part of the reexaminations proceedings for the asserted Barth patents, Rambus argued to the PTO that a synchronous system is one that, like Dan and Yano, synchronizes its data transfer operations with a clock (RX-5429C (Jacob Direct) at Q. 128; RX-4256 at 26), and an asynchronous system is one where the transmission to those devices of command and address information and data was not timed according to a clock signal (RX-27-

03990.51797/4780864.12

PUBLIC VERSION

5429C (Jacob Direct) at Q. 127; RX-4244 at 7; RX-4251 at 10; RX-4256 at 8). These statements made by Rambus during the reexamination proceedings are part of the file history of the patent and relevant when interpreting the scope of asserted claims. See, e.g., St. Clair Intellectual Prop. Consultants, Inc. v. Canon Inc., 412 Fed. Appx 270, 275-76 (Fed. Cir. 2011) (holding that [r]eexamination statements are relevant prosecution history when interpreting claims). In its response to the Commission, however, Rambus does not address whether Dan and Yano satisfy its own definition for a synchronous system. The reason for this is clear. Rambus wants to disregard its own definition because the memory systems disclosed in both Dan and Yano each synchronizes its data transfer operations with a clock (RX-5429C (Jacob Direct) at Q. 128) for the reasons set forth in Respondents response to the Commission. (RCN at 49-51; see also RX-5429C (Jacob Direct) at Qs. 492-494.) Thus, the Commission should reject

Rambus attempt to abrogate its own definitions of synchronous and asynchronous systems. Second, there is substantial evidence in both Dan and Yano that these references are synchronous. Indeed, in the case of Dan, the inventors refer to their invention as a [m]ethod and system for providing data hold time by synchronous random access memory during write operations. (RX-4262, Title Page (emphasis added).) In addition, Dan refers to clocks, clocking, and clocked through its disclosure and states that the present invention relates to a memory interface that transfers data to a data bus and writes the data to an external memory upon the clocking of a write strobe signal. (Id. at 1:01-10 (emphasis added).) Similarly, the timing diagram set forth in Figure 2 of Yano expressly shows the system clock and the timing relationship between clock and the command, address, and data information going to the memory device in a write operation. (RX-4261, Fig. 2.) Przybylski acknowledged this timing relationship during cross examination. (Tr. (Przybylski) at 2611:7-11.) 2. Rambus Arguments that Persons of Ordinary Skill in the Art Would -28-

03990.51797/4780864.12

PUBLIC VERSION

Not Seek to Combine Aspects of Synchronous and Asynchronous Systems Are Incorrect Even if the Commission were to assume arguendo that Dan and Yano are asynchronous references, which they are not, the ALJ correctly held that [t]he evidence shows that it would indeed have been obvious to combine aspects of asynchronous and synchronous art so as to achieve the inventions claimed in the Barth I patents. (ID at 212.) The Staff and Respondents set forth detailed reasons why a person of ordinary skill in the art would seek to combine aspects of asynchronous and synchronous systems. (SCN at 26-30; RCN at 52-55.) Rambus makes essentially two arguments to contrary. Each argument is without merit. (a) Jacobs Textbook Is Consistent with His Testimony in this Investigation

Rambus first attempts to draw distinctions between asynchronous and synchronous systems by quoting portions of a book published by Respondents expert Dr. Jacob in 2008. (CCN at 29-32.) Rambus arguments are irrelevant, however, as neither Respondents nor Jacob disputes that there are differences between asynchronous and synchronous systems. What

Respondents and Jacob do dispute, however, is Rambus leap from the proposition that asynchronous and synchronous systems have differences to the conclusion that [t]hese differences confirm that, as a general proposition, aspects of asynchronous systems cannot be combined with aspects [of] synchronous systems. (CCN at 31.) As Jacob testified, the simple fact that one reference discloses an asynchronous system and another reference discloses a synchronous system does not mean that it would not be obvious to one of ordinary skill in the art to combine the two. (RX-5429C (Jacob Direct) at Q. 496.) Jacob supported this position at the Hearing by explaining that [t]he earliest DRAMs actually had clocks, so SDRAMs in a sense represents a return to original principles. DRAMs went asynchronous for a long time and they came back to being synchronous with SDRAM

03990.51797/4780864.12

-29-

PUBLIC VERSION

generation. (Tr. (Jacob) at 1135:7 24.) This concept also appears in Jacobs book where he writes that [t]he earliest DRAMs (1960s to mid-1970s before de facto standardization) were often clocked . . . DRAM commands were driven by a periodic clock signal. (CX-9731.0354.) The fact that there are differences between asynchronous and synchronous systems does not mean that a person of ordinary skill in the art would not seek to combine aspects of an asynchronous system with a synchronous system. Rather, as Jacob testified, persons of ordinary skill in the art would look to both synchronous and asynchronous systems to solve problems as engineers had been doing long before the earliest filing date of the Barth patents. (RX-5429C (Jacob Direct) at Q. 496.)17 While Rambus argues that the approach of asynchronous systems is therefore incompatible with a synchronous system that demands timing precision an asynchronous system cannot provide, Rambus cites nothing in support. (CCN at 30.) In contrast, Jacob testified, [a]synchronous approaches solve problems that purely synchronous systems cannot, and vice versa. In practice, many systems represent a hybrid approach of these techniques. (RX-5429C (Jacob Direct) at Q. 496.) Again, Jacobs book fully supports this position. For example, in describing the development of SDRAM in his book, Jacob explained that the benefit of higher switching speeds found in SDRAM: applies to any interface with a source-synchronous data strobe signal, whether the interface is synchronous or asynchronous, and therefore, an asynchronous burstmode DRAM with source-synchronous data strobe could have scaled to higher switching speeds just as easily witness the February 1990 presentation of a working 100-MHz asynchronous burst-mode part from IBM, which used a
17

Rambus claim that the Barth system for the first time implemented a data strobe signal in a synchronous system to sample data is demonstrably false. (CCN at 30-31.) For example, the NextBus Specification written in 1990 disclose a synchronous system where all transitions and signal samplings are referenced to a central signal clock (RX-4265.0013) in which the data sender asserts Data Strobe (DSTB*) at the same time it places data of the bus and the receiver than captures data at the rising edge of DSTB (Id. at .0025). (See also RX-4265.0045 (stating that rising edge of DSTB* is used to sample the data).)
03990.51797/4780864.12

-30-

PUBLIC VERSION

dedicated pin to transfer the source-synchronous strobe [Kalter 1990a, b]18 (CX-9731.0358; see also CX-9731.0362 (explaining that the use of DQS in DDR SDRAMs created the implication that a free running clock could be dispensed with entirely, and the result would be something very close to IBMs toggle mode which was an asynchronous system). Jacob provided detailed testimony regarding IBMs toggle mode during the Hearing and explained that this asynchronous system used a data strobe signal called a high speed toggle signal that captured data on the rising and falling edges of the signal. (RX-5429C (Jacob Direct) at Qs. 203-205.) Therefore, both at the Hearing as well as in his book written long before his involvement in this case, Jacobs opinions were identical (i.e., (1) a data strobe can be used in either a synchronous or asynchronous interface, and (2) the use of a data strobethe only purportedly novel aspect of the asserted Barth claims19occurred long before the earliest filing date of the Barth patents). Therefore, the Commission should disregard Rambus unfounded and inaccurate attempts to use Jacobs book to contradict his testimony. (b) The ALJs Reliance on the Reexamination Findings Was Proper

Rambus also argues that the ALJ should not have considered the findings of the reexaminations of the Barth patents in rendering his opinions. Rambus clearly wants the

Commission to ignore the Barth reexaminations proceedings because the Board of Patent Appeals (BPAI) has found that all the asserted claims of the Barth patents are invalid in light of the prior art. (See RX-5356 (BPAI decision holding all claims of the 109 patent invalid); RX-5357 (BPAI decision holding all claims of the 405 patent invalid); SX-9 (Part 7) at 1059-96 (BPAI decision holding all claims of the 353 patent invalid).) In reaching these decisions, the ALJ correctly noted that the BPAI found that at the time of the filing of the original Barth patents
18

Jacobs references to Kalter 1990a, b are set forth in the References section of his book. (CX9731.0953-54.) 19 See SCN at 14 (noting that there is only one purportedly novel limitation in the asserted Barth I claimsi.e., the strobe limitation).
03990.51797/4780864.12

-31-

PUBLIC VERSION

asynchronous functions were being combined routinely with, and integrated into, synchronous DRAM systems by skilled artisans. (ID at 212 (emphasis added).) More

specifically, in reaching its decision that the asserted claims of the 405 patent are obvious in light of the prior art, the BPAI rejected the same argument that Rambus is again advancing here (i.e., that features of an asynchronous system cannot be combined with a synchronous system). (RX-5429C (Jacob Direct) at Q. 105.) Rambus position is directly contradicted by Federal Circuit case law instructing that reexamination procedure serves an important role in providing a district court with an expert view of the PTO. St. Clair, 412 Fed. Appx at 275-76.20 This is precisely how the ALJ relied on the reexamination findings (i.e., to further support of the opinions of Jacob who agreed with the BPAI that a person of ordinary skill would seek to combine aspects of asynchronous and synchronous systems). (ID at 212; RX-5429C at Q105, 129, 492-496.) Indeed, courts routinely consider reexamination proceedings in rendering their decisions.21 For these reasons, Jacob is correct in categorizing Rambus arguments based on the differences between asynchronous and synchronous systems as a red herring in this case. As the evidence demonstrates, persons of ordinary skill in the art in 1995 would seek to combine aspects of asynchronous and synchronous systemsa position Jacob set forth in his book years before he was retained in this case and that has been accepted by the ALJ, the Staff, and the BPAI. The only evidence to the contrary is Przybylskis testimony, which the ALJ wholly
20

The case cited by Rambus, In re Swanson, 540 F.3d 1368, 1377-78 (Fed. Cir. 2008), does not stand for the proposition that a court cannot rely on the PTOs reexamination findings in support of its decision as Rambus argues. (Rambus response at 32.) Rather, Swanson states that the PTO is not bound by a prior decision of a court concerning the validity of a patent. Swanson, 540 F.3d at 1378. Nothing the ALJs ID stated that the ALJ was bound to follow the PTOs findingsonly that the PTO reached the same conclusion on this issue (i.e., persons of ordinary skill would seek to combine aspects of asynchronous and synchronous systems). 21 See, e.g., Am. Piledriving Equip. v. Geoquip, Inc., 637 F.3d Cir. 1324, 1336 (Fed. Cir. 2011); Nikken, Inc. v. Robinsons-May, Inc., 51 Fed. Appx. 874, 884 (Fed. Cir. 2002); E.I. du Point de Nemours & Co. v. Phillips Petroleum Co., 849 F.2d 1430, 1439 (Fed. Cir. 1988).

03990.51797/4780864.12

-32-

PUBLIC VERSION

rejected based on his observations of Przybylski at the Hearing. (ID at 305-309.)22 C. Harriman Evidences Publication of the NextBus Specification (Issue 2c)

In its Response to the Commissions Notice of Review, Rambus spends little time addressing the Commissions question regarding whether the Harriman patent evidences publication of the Next Bus specification in view of the fact that NeXT is the assignee of the Harriman patent. Rather Rambus makes the same arguments the Respondents already addressed in their response to Rambus petition for review. (RRPFR at 12-15.) As a result, Rambus arguments require little response here. First, Rambus argues that there is no way of knowing the content of the NextBus Specification or whether it is the same as the Next Document in question. (CCN at 36.) When asked this question at trial, Jacob answered that [t]he diagrams for the signals are the same, substantially the same in the patent and specification. The names of the signals are the same. The description of the mechanism is the same. It is the same thing. (Tr. (Jacob) at 1154:22 1155:7 (emphasis added); see also Tr. (Jacob) at 1165:22-1166:9.) Moreover, the prominent features of NextBus, such as the data strobe signal and double data rate are disclosed in the Harriman patent. (See, e.g., RX-4266.0019 at 15:64-66.) In fact, both the NextBus Specification and Harriman state that Next Bus is a superset of NuBus, which is defined by IEEE standard 1196. The NextBus has different clocks, data transfer rate and electrical interface than NuBus.

Rambus remaining arguments should also be disregarded. First, Rambus attempts to distinguish Jacobs reliance on evidence from Rambus own witnesses concerning the combination of synchronous and asynchronous systems are without merit. (CCN at 35-36.) Jacob simply cited the testimony of Stark and Barth to provide further examples that persons of skill in the art would routinely seek to combine aspects of both types of systems. This is precisely what these examples show. (RX-5429C (Jacob Direct) at Q. 496.) Second, Rambus attempt to argue that JEDECs use of a data strobe in 1999 supports its position that the inventions claimed in the Barth patents were novel misses the point. (CCN at 36.) Jacob simply was providing an example of the use of asynchronous strobe in a synchronous system was common. Indeed, the ALJ found DQS signal in the JEDEC DDR Specifications was identical the DSTB* signal in the NextBus Specification developed in 1990 over five years before the original filing date of the Barth patents. (ID at 112.)
03990.51797/4780864.12

22

-33-

PUBLIC VERSION

(RX-4266.0019 at col. 15:50-53; RX-4265.0013.) Indeed, there is more than enough evidence to support the ALJs finding that the Next Bus Specification incorporated in Harriman is the same Next Bus Specification relied on by Respondents in this Investigation. (ID at 104-06; see also SCN at 37-39.) Second, Rambus relies on Przybylskis testimony that the Next Bus Specification may have been provided under confidentiality agreements. (CX-10765C at Q. 232.) Pryzbylski, however, cites to no evidence whatsoever to support this position. This is particularly telling where all the evidence in the NextBus Specification and Harriman demonstrate otherwise. For example, NeXT disclosed the prominent features of NextBus in the Harriman patenta public document issued well over a year prior to earliest filing date of the Barth patents. Moreover, there is no confidentiality language on the NextBus Specification, which courts rely on in deciding questions of public availability. See, e.g,. In re Natures Remedies, Ltd., 315 Fed. Appx 300, 305 (Fed. Cir. 2009).23 Finally, Rambus argument based the fact that the website listed on the last page of the NextBus Specification was not available until 2005 is irrelevant. As Jacob explained, the page listing the website is in a different font and different size, different everything. It is pretty clear that this [page] is stapled on at the end after the fact, (Tr. at 1167:8-17). When this website was created or the page listing the website was first published is irrelevant to when NextBus was publicly available. Indeed, Harriman, which was filed in 1991, states that the NextBus

Specification was available from NeXT Computer. (RX-4266 at 15:32-60; SRPFR at 19.)
23

Rambus reliance on Kyocera Wireless Corp. v. ITC, 545 F.3d 1340, 1351-52 (Fed. Cir. 2008) is easily distinguishable. (RCN at 37.) Specifically, in Kyocera, the Court rejected defendants attempt to claim that ten separate documents were a single document based on a reference to a single standard. Kyocera, 545 F.3d at 1351-52. This is far from the case here where Harriman specifically refers to the NextBus Specification and states where the reference is available (i.e., from Next Computer). (RX-4266 at 15:4360; see also Cook Biotech Inc. v. Acell, Inc., 460 F.3d 1365, 1376 (Fed. Cir. 2006) (finding sufficient incorporation by reference by identifying the procedure to be incorporated and the patent disclosing it); Advanced Display Sys., Inc. v. Kent State Univ., 212 F.3d 1272, 1282 (Fed. Cir. 2000).
03990.51797/4780864.12

-34-

PUBLIC VERSION

D.

Respondents Have Demonstrated that SyncLink Is Prior Art (Issue 2d)

Respondents set forth detailed evidence in its Response to the Commissions Notice of Review demonstrating the public availability of SyncLink prior to the alleged October 19, 1995 priority date of the asserted Barth patents. (RCN at 59-63.) Rambus arguments to the contrary fail to undermine any of Respondents positions. First, Rambus arguments that SyncLink had limitations on its dissemination and that it was in draft form are irrelevant. As set forth in detail in Respondents response, SyncLink itself demonstrates that the IEEE made the SyncLink reference accessible to members of the interested public for the purpose of standardization and developing a national position, and was open to comments from any interested party. (RCN at 60-61.) Moreover, Rambus cites no

authority that a document must be in final form to be invalidating prior art. The IEEE published SyncLink in draft form to the interested public to develop consensus for a national standard. Indeed, to accomplish this, the IEEE had to make SyncLink available to the interested public. (See, e.g., RX-2048.0001 (containing meeting minutes from a May 24, 1995 JEDEC meeting discussing SyncLink with representatives from over 30 companies in the industry to further develop the standardization of SyncLink).) Second, Rambus attempts to mitigate the evidence of its own possession of SyncLink materials to the issue of whether SyncLink was publicly available are simply not credible. Respondents set forth an abundance of evidence showing that Rambus, as a member of the interested public, had ready access to SyncLink standards documents prior to the filing of the first Barth patent on October 19, 1995. (RCN at 61-63.) In response, Rambus argues that no one with personal knowledge testified that the SyncLink document was publicly available. (CCN at 41.) Yet Rambus again ignores its own evidence. In particular, Crisp, an employee of Rambus who attended the SyncLink meetings in 1995, testified that SyncLink documents were -35-

03990.51797/4780864.12

PUBLIC VERSION

available publicly through an FTP site during the time he attended SyncLink meetings. (JX030.0029 at 85:1-21; 97:3-17 (emphasis added); see also RX-2050C.0001, .0005 (an email from Crisp dated May 24, 1995 acknowledging he had attended all the Synchlink [sic] meetings).) When questioned about how he obtained the SyncLink document he posted outside his cube on September 29, 1993 (see RX-4287C), Crisp replied that it was [j]ust a publicly-available document that I downloaded from the Web. (JX-030C.0028 at 83:20-84:25 (emphasis added).) Thus, Rambus cannot credibly argue that SyncLink was not publicly available prior to the claimed priority date of the Barth patents when every piece of evidence, including testimony from its own employee, demonstrates otherwise. V. SDR PRODUCTS DO NOT INFRINGE THE BARTH PATENTS (ISSUE 3D) STMicros Cartesio1, Hydra, and Yeti 3.1 products do not infringe when STMicros customer connects to SDR memory instead of DDR memory. (RCN at 63-64.) Rambus and the Staff do not contend otherwise. (CCN at 48-50, SCN at 47-48.) The record shows that manufactures devices with Cartesio1 connected to SDR memory. The record also shows that Seagate makes disk drives with Yeti connected to SDR memory. (RX-6273C (Siraz) at Q45, 47-50.) Rambus contends that these STMicro products have no non-infringing uses because once configured to control DDR memory, they do not operate with SDR memory. (CCN at 48-50.) Rambus misses the point, because when STMicro ships a Cartesio1, Hydra, or Yeti 3.1, it is fully capable of being connected to non-infringing SDR memory. It is STMicros customer, not STMicro, who decides whether to connect to SDR or DDR memory. (See, e.g., RX-6031C at 24, SDR or DDR mode is selected by an input from surrounding circuitry.) Rambus argues that the SDR interface is a separate and distinct feature, but the evidence is to the contrary. Hydra has an SDRAM interface for SDR and DDR with shared data,

03990.51797/4780864.12

-36-

PUBLIC VERSION

address, and control signals. (RX-6031C at 25.) Yeti has a memory interface for Synchronous DRAM that can be either Single or Dual Data Rate. (RX-6040C at 9; see also, id. at 18 (shared memory interface ports for SDR and DDR); RX-6035C at 412.) The SDR and DDR interfaces are not separate and distinct because they share data, address, and control circuitry. The record thus establishes that STMicros Cartesio1, Hydra, and Yeti 3.1 have substantial non-infringing usesinterfacing to admittedly non-infringing SDR memoryand thus do not infringe under 35 U.S.C. 271(c). VI. THE ALJ CORRECTLY FOUND THE BARTH PATENTS UNENFORCEABLE DUE TO RAMBUS UNCLEAN HANDS (ISSUE 4) A. 4(a): Rambus Should be Prevented from Challenging the 661 Determinations of When It Had a Duty to Preserve Documents and Its Bad Faith

Neither Rambus nor the Staff raises any challenge to three of the required elements of collateral estoppel, effectively conceding that application of the doctrine hinges on just one. (CCN at 51-53; SCN at 48-55.) Rambus and the Staff dispute whether the ALJs determinations were necessary to the judgment in the 661 Investigation. (Id.) In their view, no matter how much effort went into litigating and deciding them, the resolution of these issues was unnecessary because the 661 Respondents did not win on unclean hands. But no rule as stringent as this exists: a determination need not be so integral to a judgment that the judgment would not stand without it. Shell Petroleum, Inc. v. U.S., 319 F.3d 1334, 1339 (Fed. Cir. 2003) (citing Mothers Rest., Inc. v. Mamas Pizza, Inc., 723 F.2d 1566, 1571 (Fed. Cir. 1983))24; see, e.g., In the Matter of Certain Ground Fault Circuit Interrupters, Inv. No. 337-TA-739, 2011 WL 7168646, at *37-40 (Dec. 20, 2011) (precluding complainant from challenging the materiality of a prior art reference where the Federal Circuit had vacated
The Staff argues that Mamas Pizza does not support Respondents position because the precluded party lost in the prior action and the rejection of the claim rested on determination of the issue that party sought to relitigate. (SCN at 52-53.) Neither of these facts, however, figured into the Federal Circuits analysis of whether the issue was necessary. See Mamas Pizza, Inc., 723 F.2d at 1569-71.
03990.51797/4780864.12
24

-37-

PUBLIC VERSION

the lower courts summary judgment of inequitable conduct but expressly affirmed its materiality finding). This is because, taken to its logical end, a contrary rule would gut the doctrine. Rather, as explained by the Federal Circuit, the necessary requirement serves merely to prevent the incidental or collateral determination of a nonessential issue from precluding reconsideration of that issue in later litigation. In re Freeman, 30 F.2d 1459, 1466 (Fed. Cir. 1994); Shell Petroleum, 319 F.3d at 1339 (same). Consistent with this purpose, collateral estoppel is appropriate where an issue has already been subject to a considered determination. See RESTATEMENT (SECOND) OF JUDGMENTS 27, cmt. h (1982).25 Neither Rambus nor the Staff deny that the ALJ made considered determinations of the spoliation issues in the 661 Investigation. A lengthy discussion of these issues spans dozens of pages in the I.D. (See, e.g., 661 ID at 92-116.) Resolving the spoliation issues without due consideration was never an option since the ALJ had to thoroughly address those issues under the law. 19 CFR 210.42(d). (The initial determination shall include: an opinion stating findings . . . and conclusions and the reasons or bases therefor necessary for the disposition of all material issues of fact, law, or discretion presented in the record . . .) All of the spoliation findings were adopted without modification, confirming their legal sufficiency. (See 661

Commn Op. at 2.) The ALJs rulings thus have none of the characteristics of dicta that would render them unnecessary to the judgment. Rambus attempts to erect another barrier to collateral estoppel based on the right to appeal. (Id. at 52.) Yet, as the Supreme Court case cited by Rambus makes clear, no per se rule based on appellate review exists. Standefer v. U.S., 447 U.S. 10, 23 (1980) ([T]his is not to suggest that availability of appellate review is always an essential predicate to estoppel . . . .).
The hypotheticals the Staff cites from the Restatement are inapposite. (SCN at 51-52.) Both are classic examples of dicta, in which one of two alternate defenses is rejected. The two spoliation issues here, by contrast, were each essential to the unclean hands claim, not offered as alternatives.
03990.51797/4780864.12
25

-38-

PUBLIC VERSION

Moreover, any such rule would not apply here because Rambus had an opportunity to appeal the ALJs findingsand did. Indeed, Rambus not only had the right to petition for review of the I.D., in the appeal of the 661 Investigation Rambus challenged the entirety of the ALJs anticipation of litigation analysis, requesting that it should, at the very least, be vacated by the Federal Circuit. (Brief of Rambus Inc. dated May 9, 2011, at 54-57.) Rambus argues that this should be ignored because it could not itself directly appeal the ALJs findings. (CCN at 52 n.14 (emphasis added).) But this is a distinction without a difference, and one that finds no support in Rambus cited authorities. The doctrine of stare decisis should also apply to the ALJs determinations because the same law is being applied to the same operative facts. (RCN at 80-81.) The only changes to the law in the spoliation context favor Respondents and do not affect the ALJs analysis of these two issues.26 (Id.) Moreover, while the record on spoliation has improved for Respondents, all of the facts that formed the basis for the ALJs determinations on anticipation of litigation and bad faith are, as Rambus admits, present here. (CCN at 54.) Even if stare decisis does not apply, its underlying rationale compels the same result as if it did. Stare decisis exists to ensure consistent decision-making through the practice of following precedent and to set expectations for litigants. Moragne v. States Marine Lines, Inc., 398 U.S. 375, 403 (1970) (listing weighty considerations that underlie stare decisis). These laudable goals can be served if Rambus challenges to the prior determinations are barred. B. 4(b): Respondents in this Investigation Presented Additional Factual Evidence on Prejudice Through Live Testimony and Videotaped Depositions 1. The Record in This Investigation Reflects Substantial New Evidence on the Issue of Prejudice Adduced by Respondents at the Hearing

Rather than merely recycle past evidence as Rambus contends, Respondents were able
By contrast, stare decisis would not apply to the prejudice determinations because the law and facts on prejudice have changed since the 661 Investigation, both in Respondents favor.
03990.51797/4780864.12
26

-39-

PUBLIC VERSION

to improve the already extensive record on spoliation27 from the 661 Investigation by introducing substantial additional evidence of prejudice. (CCN at 53.) At the hearing, for instance,

Respondents elicited new, unique testimony on prejudice live from numerous witnesses. (RCN at 81-87.) Some of these witnesses, including former CEO Geoffrey Tate, in-house counsel Joel Karp, and outside patent counsel Lester Vincent, none of whom had not been called to testify in the 661 Investigation. (Id. at 81-83, 85.) Others, such as inventor Richard Barth, Craig Hampel, and Jose Moniz, provided new testimony at the hearing. (Id. at 82-85.) Respondents also adduced new evidence on prejudice by designating testimony from prior depositions that had not been designed in the 661 Investigationfor example, from Neil Steinberg, Rambus former General Patent Counsel. (Id. at 86.) Likewise, Respondents presented new trial testimony from another litigation in California that took place after the 661 hearing. The fact that discovery took place in prior litigations does not, as Rambus contends, somehow foreclose the possibility that Respondents revealed new facts regarding Rambus spoliation or exposed additional untruths. (CCN at 54.) Notably, most of the prior litigation concerned the Farmwald-Horowitz patents, and thus the focus only recently shifted to the Barth patents. Moreover, as explained, Respondents adduced new evidence through the testimony of various witnesses at the hearing that documents Rambus has never produced in any litigation such as patent conceptual trees and emails from Barth about drafting claims to cover JEDEC technologiesexisted but were destroyed in violation of Rambus duty to preserve. (RCN at 8187; see ID at 300-05.) Despite the many years of litigation over Rambus spoliation, this
Despite acknowledging that Question 4(b) is limited to distinctions between the two investigations, the Staff faults Respondents for relying on documents from the 661 Investigation in their petition for review. (SCN at 55-58.) Besides being nonresponsive, this argument ignores that, under the new prejudice framework, the burden shifted to Rambus to show lack of prejudice. Micron II, 645 F.3d at 1328. Moreover, a review of the entries on the chart indicates that not all of the documents were admitted in the 661 Investigation. Some were only produced in discovery but not admitted at the hearing; others were withdrawn; and still others do not match the descriptions given. (SCN at 56-57 (listing, e.g., RX2047C, RX-2111C, RX-2524C, RX-2463C, RX-2577C, RX-2702C, RX-2709C; RX-3909, RX-3918C).)
03990.51797/4780864.12
27

-40-

PUBLIC VERSION

testimonial evidence reveals details not previously known about what was destroyed. (Id.) 2. Respondents Called Rambus Witnesses to Develop Additional Facts, Assess Their Credibility, and to Impeach with Prior Testimony

Rambus presumes that Respondents called witnesses to testify live only to highlight selected excerpts of past testimony rather than to elicit new evidence. (CCN at 54.) On the contrary, the records shows these witnesses were called to develop new testimony on substantive issues relevant to the Investigation (e.g., Respondents new defenses).28 (RCN at 81-87.) The hearing also presented the first opportunity for the ALJ to observe the demeanor of witnesses not called in the 661 Investigation. Live cross-examination allowed for those witnesses to be

confronted with contradictory or suspect prior testimony. Their responses provided insight into whether other facts to which the witnesses had testified should be believed. The ALJ found multiple witnesses testified dishonestly or displayed a selective memory on the stand, casting doubt on not just their testimony in this investigation but in previous litigations. (See, e.g., ID at 270-305, 309-11, 313.) Rambus cannot and does not deny this evidence is new. 3. The New Evidence in the Record Does Not Support Rambus Position That It Destroyed Nothing Helpful To Respondents Defenses

Unwilling to yield on any point, after taking the position that there is no new evidence in the record, Rambus simultaneously argues that any new evidence exclusively supports its positions. (CCN at 55-58.) This idealized view of the factual record fails to reflect reality. Moreover, it ignores the ALJs dim view of the demeanor of Rambus witnesses on the stand. (a) Respondents Presented New Testimonial Evidence of Spoliation From Rambus Witnesses at the Hearing

During the hearing, Respondents obtained or presented new testimony that spoliation had occurred at all levels of Rambus corporate structureincluding the very top. Testimony

In its chart showing that witnesses have testified extensively in the past, including on Rambus adoption and implementation of its document retention policy, Rambus includes depositions taken in this Investigation from July 2011. (CCN at 55.)
03990.51797/4780864.12

28

-41-

PUBLIC VERSION

regarding spoliation by executives was elicited from spoliation ringleader Joel Karp, Rambus VP of Intellectual Property, and even Rambus co-founder Mark Horowitz. (RCN at 85.) Karp testified that he went into the office of Rambus CEO at the time, Geoff Tate, to help him select documents to shred. (Id.) Respondents offered similar testimony from Horowitz that someone went into his office in 1998 and selected documents to shred. (Id.) Tate himself confirmed the importance of email at Rambus, stating that in emails employees could talk about anything, including inventions, patent, competitive products or standards, and JEDEC. (Id. at 85.) Respondents also presented new testimony regarding the destruction of lead inventor Richard Barths documentsdocuments which would have been highly relevant to Respondents defenses. (Id. at 82-85; ID at 300-05.) Barth testified, for example, that none of the five conceptual patent trees produced by Rambus included any of the handwritten notes he made on those documents from time to time. (RCN at 83.) He also testified that none of the trees had any references to the original Barth patent (denoted P048) as would be expected. (Id. at 83.) Barth further admitted that when he quit Rambus, he left behind tens of thousands of emails, all of which would have been subject to the retention policy, and that he had no idea what became of those emails. (Id. at 84.) It was also revealed during Karps cross examination that Karp had sweated bullets searching for evidence of a conception date for one of Barths inventions on a backup tape that was to be destroyed. (Id. at 84-85.) (b) Respondents Established that Rambus Outside Counsel Purged Patent Files Before Transferring Those Files to Rambus

The live testimony of Lester Vincent, Rambus outside patent counsel, did not, as Rambus and the Staff argue, demonstrate a lack of prejudice. (CCN at 55-56; SCN at 59-62.) Far from it. Vincent, who had been instructed by Karp to apply the document retention policy to the patent files in April 1999, confirmed that he cleaned files for the Barth patent applications. (RCN at 82; RIB 238-240.) His testimony established that documents expected to be in the
03990.51797/4780864.12

-42-

PUBLIC VERSION

Barth files were missingand thus likely purged. (RCN at 82-83.) For example, he testified that the files did not contain any notes from discussions with inventors despite his habit of notetaking and records showing he billed for hours of meetings with Barth. (Id.) Similarly, the files did not contain any draft applications and there was no correspondence in the files for other Barth patent applications. (Id. at 83.) He further testified the files did not contain floppy disks even though the purging chart indicates the files once contained such disks. (Id. at 82-83.) Rambus attempts to draw a distinction between the Barth patent files and the asserted Barth patent files, implying that only destroying documents kept in the latter would run afoul of its duty to preserve documents relevant to this litigation. Yet documents relevant to

Respondents defenses could have been in either set of filesbut were most likely in both. See, e.g., Microsoft Corp. v. Multi-Tech. Sys., Inc., 357 F.3d 1340 (Fed. Cir. 2004) (finding statements made during prosecution of sibling patent relevant to asserted patent). Contrary to Rambus assertions, the testimony of Jose Moniz did nothing to neutralize the Barth patent file cleaning. (CCN at 57.) As the ALJ pointed out, Moniz was not even at Rambus when the purging of Blakelys files occurred and he had no knowledge of whether any Rambus employee had cleaned Rambus Barth patent filesincluding those of the asserted patentsprior to his arrival in late 1999. (ID at 311.) Moreover, his prior testimony can hardly be considered undisputed; in fact, Moniz himself admitted at the hearing that his testimony in the 661 Investigation included a significant mixup. (RCN at 82.) Under cross-examination, Moniz revealed that his testimony in the 661 Investigation that Vincent transferred all the Barth patent files to Rambus before the cleaning was wrong. (Id.) Moniz had little choice: Vincent, who took the stand before Moniz on the same day of the hearing, testified that the transfer did not occur until mid-2000, well after the cleaning in 1999. (Id.) Other attempts by Rambus and the Staff to explain away the purging of the patent files -43-

03990.51797/4780864.12

PUBLIC VERSION

fare no better. Rambus asserts with certainty that contents of the missing disks were entirely duplicative. (CCN at 56.) But it has never proven thisnor could it given that there is no record of what was on the discs or what was transferred to Rambus. (Tr. (Vincent) at 2245:8-17.) The Staff claims the copies of those contents could largely be obtained from other sources, conceding that other documents contained on the discs may have been lost. (SCN at 60.) Likewise, nothing can be gleaned about the contents of the patent files from the pace of prosecution. (CCN at 57-58; SCN at 61.) Indeed, prosecution may have proceeded without incident precisely because Rambus knowingly withheld documents like SyncLink. Any notes in the files regarding prior art not disclosed to the PTO would have been highly relevant to Respondents defenses. Finally, Rambus and the Staff argue that the patent files would not have been discoverable. (CCN at 58; SCN at 60.) But documents do not attain privileged status merely because of their storage in prosecution files. In any event, Rambus privilege had already been pierced on the basis of fraud. (RIB at 119-20 (citing 2001 order).) 4. Respondents Developed New Facts In Proving Up Their New Defenses

Rambus argues that Respondents new defenses in this Investigationwritten description, prosecution laches, and inequitable conduct to name a fewhad no impact on the factual content of the record. (CCN at 58.) To prove the unique elements of these new defenses, however, Respondents had to seek new facts through discovery and adduce new evidence through documents29 and testimony at the hearing. (RCN at 87.) As a result, the record here overlaps with, but goes well beyond, the record in the 661 Investigation. As but one example of how the record differed, Respondents had to prove Rambus had an
29

Rambus contention that the existence or nonexistence of Rambus documents has no bearing on noninfringement or validity is simply wrong. (CCN at 58.) An internal document detailing why Rambus believed an accused technology did not use technology claimed in the Barth patents would bear on non-infringement. Similarly, a document casting doubt on patentability would bear on invalidity. Other documents, including those bearing on Respondents inequitable conduct and laches defenses, would also have been relevant to Respondents defenses. (RIB at 256-58; RRB at 125-26.)
03990.51797/4780864.12

-44-

PUBLIC VERSION

intent to deceive the PTO and that those two references are material to prove their inequitable conduct defense based on SyncLink and RamLink, a defense which was not raised in the 661 Investigation. (RIB at 209.) To support this defense, Respondents presented new evidence regarding Rambus awareness of SyncLink and RamLink, and its view of the materiality of those references as well as new arguments on these issues. (RIB at 211-16.) New testimony regarding SyncLink and RamLink came from, among others, Richard Barth and Richard Crisp, Rambus JEDEC representative. Testifying live at the hearing, Barth admitted to being aware of the SyncLink reference and its connection to JEDEC, stating that Crisp provided him with copies of JEDEC meeting minutes discussing SyncLink. (Tr. at 407:17408:15.) He also admitted to receiving an email from Crisp during prosecution of a Barth parent application stating: Well, if it is possible to salvage anything that helps us get a claim to shoot SynchLink in the head, we should do it . . . . (Tr. at 405:15-406:13.) This email was apparently not admittedand certainly not emphasizedin the 661 Investigation. (RX-4288C.) Under questioning directly from the ALJ at the 753 hearing, Barth testified live for the time that he must have had the Synchlink document prior to actually filing the Barth applications. (Tr. at 476:14-16.) Barth could not explain why he decided not to disclose SyncLink, and stated for the first time that if it were possible, he would want to revisit the decision not to disclose it. (Tr. at 450:9-12; 474:15-22.) Based in part on these new admissions, the ALJ concluded that

documents relevant to drafting of the Barth patentsincluding to claim competing technology such as SyncLinkwere missing. (ID at 260, 305, 313). To further support their inequitable conduct claim, Respondents designated testimony from Crisps deposition taken in this Investigation, as well as other sources that were not part of the 661 record. (JX-030C.) At his deposition taken in this Investigation, Crisp admitted he was a member and attended meetings of the IEEE RamLink and SyncLink working groups and -45-

03990.51797/4780864.12

PUBLIC VERSION

therefore had access until mid-1995 to RamLink and SyncLink specifications. (JX-030C.002836.) Respondents also designated excerpts from Crisps trial testimony on August 22, 2011, which was given after the 661 hearing and in which he testified he expected all the documents from his office that he loaded into burlaps sack would be destroyed. (JX-030C.0007-11.) None of this testimony existed during the 661 Investigation. Finally, Rambus and the Staff contend that the existence of SyncLink and RamLink documents shows that Respondents suffered no prejudice. (CCN at 59; SCRN at 58-59.) But keeping some documents does not mean others were not eliminated. Indeed, Karp instructed Rambus employees to save helpful documents and to destroy harmful ones, such as those casting doubt on patentability. (RIB at 236.) Any documents bearing on inequitable conduct surely would have cast doubt on the patentability of the Barth Patents and thus been targeted for destruction. Moreover, whether or not SyncLink or RamLink were publicly available says nothing about Rambus view of those references or decisions to withhold them from the PTO. VII. THE ALJ ERRED IN NOT FINDING THE BARTH PATENTS UNENFORCEABLE DUE TO RAMBUS INEQUITABLE CONDUCT (ISSUE 5) A. 5(a): Materiality of RX-4270C (SyncLink) Rambus discussion of Issue 5(a) centers on Rambus mistaken assertion that Respondents failed to take into consideration the but-for test for materiality under the Federal Circuits Therasense decision.30 But, to the contrary, as Respondents pointed out in their initial post-hearing brief: The standard for materiality in the context of inequitable conduct is but-for materiality. Therasense, Inc. v. Becton, Dickinson & Comp., No. 2008-1511, at 27-28 (Fed. Cir. 2011) (en banc). [I]n assessing the materiality of a withheld reference, the court must determine whether the PTO would have allowed the claim if it had been aware of the undisclosed reference. Id. Courts should adopt
30

Rambus mistakenly reads Respondents brief as contending that the strobe disclosed in RX-4270C is the strobe in the asserted Barth claims. (CCN at 62.) As set forth in their response to the Commissions Notice of Review, Respondents point to the flags as disclosing the strobe element of the asserted Barth claims. (RCN at 67-69.) Consequently, there has been no change in position.
03990.51797/4780864.12

-46-

PUBLIC VERSION

USPTO standardsnot those of courts in infringement suitsin making this materiality determination: the court should apply the preponderance of the evidence standard and give claims their broadest possible construction. Id. (RIB at 209.) Furthermore, Respondents noted in their reply to Complainants initial posthearing brief that Rambus refused to acknowledge that under Therasense the test for materiality is a preponderance of evidence standard giving the claims their broadest possible construction. (RRB at 95.) Respondents contingent petition for review did not revisit this discussion because the only matters put at issue in the ID were (1) whether SyncLink was cumulative of the Farmwald prior art; and (2) whether Respondents had presented sufficient evidence of intent. (RPFR at 32.) The Commission, by contrast, in its Notice of Review asks whether the SyncLink reference relied on by Dr. JacobRX-4270Cis material to the patentability of any or all asserted Barth claims. Respondents have answered this question by pointing out that the specific documentRX-4270Cis a sufficient disclosure on its own to anticipate all asserted claims of the 353 Patent and claims 1, 20, and 24 of the 109 Patent.31 (RCN at 65.) In reaching his conclusion that SyncLink anticipated these claims, Jacob relied and referenced solely RX4270C. (Id.) Therefore, under the but-for test for materiality established in Therasense giving the claims their broadest reasonable constructionRespondents have proven by a preponderance of the evidence that RX-4270C is material to the patentability of at least all asserted claims of the 353 Patent and claims 1, 20, and 24 of the 109 Patent. Rambus also argues that the flags cannot be the strobe in the Barth claims because the 1/0 pairs are associated with all intermediate bytesessentially arguing that the same two
31

Respondents presume that the Commissions question may have been prompted by Respondents posthearing briefing on inequitable conduct which points to both SyncLink and RamLink. Thus, Respondents understand the Commissions Issue 5(a) to be directed towards the specific question of whether RX4270C by itself anticipates any asserted claim or renders any asserted claim obviousthereby rendering that specific document a material reference.
03990.51797/4780864.12

-47-

PUBLIC VERSION

bits of information are used to designate data and address information. (CCN at 64.) This argument, however must fail because in the Barth specification the strobe is a single bit of data and has identical characteristics to both the terminate signal and the wakup signal. (See JX05.0030 [109 Patent at col. 14, lines 11-13] [The terminate signal for transaction 0 and the wakeup signal for transaction 2 both have identical characteristics to strobe signals.]) Thus, the fact that identical signals in SyncLink may be used to designate different actions or events is a red herring, which is precisely what Jacob pointed out during his cross-examination at the hearing. (RPFR at 31.)32 Finally, Rambus arguesconsistent with the ALJthat SyncLinks flags are equivalent to the AddrValid bit in Farmwald such that SyncLink is merely cumulative of art already considered and rejected by the PTO. (CCN at 65-66.) Rambus argument fails for at least two reasons: (1) Farmwald has been found by the PTO to anticipate the non-strobe embodiment of the 109 patent whereas SyncLink teaches one of the strobe embodiments; and (2) during prosecution of the 109 patent, the applicant successfully overcame a rejection based on Farmwald by arguing that the AddrValid signal was not the claimed signal because it went to an intermediate transceiver device, not the claimed memory device, whereas SyncLink does not use an intermediate transceiver device. Each of these arguments have already been fully addressed in prior briefing. (RCN at 67-69.) B. 5(b): Breadth of Signal and Strobe Signal Claim Elements

Rambus first argument is that Respondents should be precluded from arguing that signal be construed more broadly than strobe signal. (CCN at 66-67.) First, the

Commission did not ask whether the Respondents could raise the issue; rather, the Commission
32

Rambus argument the flags in SyncLink cannot be the strobe because in certain circumstances there is no data is also a red herring. (CCN at 65.) Nothing in the claims requires that each time a strobe signal is issued it must initiate sampling. All that is required to anticipate the claim is that the strobe be issued at least once to initiate sampling by signaling to the memory device that valid data is on the bus.
03990.51797/4780864.12

-48-

PUBLIC VERSION

simply asked whether the PTOs construction of signal is broader than the ALJs construction of strobe signal. (NOR at 4.) Respondents answered this question in the affirmative.33 (RCN at 65-67.) Thus, Rambus waiver argument is beyond the scope of the issues raised in the Commissions Notice of Review and should therefore be disregarded. (NOR at 3 [the parties are asked to respond only to the questions enumerated below].) Rambus next argues that the PTOs construction of signal is inconsistent with the Commissions findings in the 661 Investigation. (CCN at 67-70.) Here again, Rambus

addresses an issue not raised by the Commissions Notice of Review.34 The Commission asked whether the PTOs construction of signal is broader than the ALJs construction of strobe signalnot whether the PTOs construction was consistent with findings in the 661 nor whether Respondents should be precluded from arguing for such a construction. (NOR at 4.) Lastly, Rambus asserts that SyncLink is not material even under the PTOs construction of signal in the 109 claims. (CCN at 70-71.) This argument too should be disregarded as it sheds no light on the simple question asked by the Commission: whether the PTOs construction of signal is broader than the ALJs construction of strobe signal.35 C. 5(c): Effect of PTOs Construction on Whether SyncLink is Cumulative

Here, as before, Rambus mistakenly reads Respondents contingent petition for review as contending that the strobe disclosed in RX-4270C is the strobe in the asserted Barth claims.
33

As explained below and in prior briefing, Respondents contend that SyncLink anticipates certain strobe signal embodiments of the asserted Barth patent claims. Because the strobe signal embodiment is a species of the broader signal embodiments in the 109 patent, there is no inconsistency in Respondents contentions. The fact that the signal element broadly covers both strobe-signal and non-strobe-signal embodiments is immaterial to Respondents contentions with respect to SyncLink. Moreover, with respect to Farmwald 037, Respondents did argue at the hearing and in post-hearing briefing that signal was a broader term that included strobe and non-strobe embodiments. (RIB at 7173.) Thus, Rambus waiver argument should be rejected. 34 This section of Rambus response has little to do with an inconsistency with the Commissions findings in the 661 and is really just a rehash of its opening waiver argument. Regardless, the argument is beyond the scope of the issues raised in the Commissions Notice of Review and should be disregarded. 35 The merits of Rambus argument have already been addressed. (RCN at 67-70.)
03990.51797/4780864.12

-49-

PUBLIC VERSION

(CCN at 71-72.) As set forth in their response, Respondents point to the flags as disclosing the strobe element of the asserted Barth claims. (RCN at 67-69.) Consequently, there has been no change in position and Rambus assertion to the contrary is in error. As Respondents have consistently argued, SyncLink is material and non-cumulative under either the PTOs broad construction of signal or under the ALJs broad construction of strobe signal. (RCN at 65-69.) This is because (1) Farmwald has been found by the PTO to anticipate the non-strobe embodiment of the 109 patent whereas SyncLink teaches one of the strobe embodiments; and (2) during prosecution of the 109 patent, the applicant successfully overcame a rejection based on Farmwald by arguing that the AddrValid signal was not the claimed signal because it went to an intermediate transceiver device, not the claimed memory device, whereas SyncLink does not use an intermediate transceiver device. Each of these arguments have already been fully addressed in prior briefing. (RCN at 67-69.) D. 5(d): Scope of Unenforceability Finding

The Commission asks whether the 353 and 405 patent claims should be held unenforceable if the Commission were to find that the 109 claims are uneforceable due to inequitable conduct. (NOR at 4. 36 ) Respondents believe the answer is yes because, as

Respondents have shown, SyncLink anticipates all asserted claims of the 353 patent and renders obvious the claims of the 405 patent. (RCN at 69-70.) Rambus first argument is that the Respondents waived any argument that SyncLink is material to the 405 patent. (CCN at 72.) The Commission, however, did not ask whether any party had waived an argument as to the unenforceability of the 405 patent, and Rambus

Respondents presume this question is in the context of a finding that Rambus intentionally and knowingly failed to disclose SyncLink (RX-4270C) during prosecution of the 109 patent.
03990.51797/4780864.12

36

-50-

PUBLIC VERSION

argument should therefore be disregarded.37 Rambus next asserts that SyncLink is not material to either the 353 or 405 patent. (CCN at 72.) Respondents have shown, however, that SyncLink anticipates claims 11-13 of the 353 patent. (RCN at 69-70.) Moreover, Rambus did not dispute that Farmwald discloses all elements other than the signal and strobe signal elements. (ID at 121 [It is important to note that the only elements of claims 1, 2, 4, 5, 12, 13, 20, 21, and 24 of the 109 patent that Przybylski argues are not disclosed in Farmwald 037 are the elements concerning the signal (or code) that indicates when the memory is to begin sampling.].) Thus, even if SyncLink does not anticipate the asserted claims of the 405 patent, its disclosure of the preferred, strobe embodiment (identified as the indirect control embodiment by the ALJ) renders it a material reference to the 405 patent as its combination with Farmwald renders the asserted claims of the 405 patent obvious.38 Lastly, Rambus argues that there was no intent to deceive the patent office. (CCN at 7274.) But this is ground that was well covered in earlier briefing (RPFR at 34-36) and beyond the scope of the Commissions Notice of Review. 39 Rambus assertion herethat the JEDEC meeting minutes related to SyncLink were submitted during prosecution of the 353 and 405 patents, and thus, there can be no finding of intent to deceive as to those patentsis irrelevant to the question presented by the Commission. There is no dispute that RX-4270C was not

37

Moreover, contrary to Rambus assertions, Respondents did contend at the hearing and in post-trial briefing that SyncLink was material to all asserted Barth patent claims. (RIB at 216 [This establishes that RamLink and SyncLink are material to the patentability and prosecution of all of the asserted Barth claims.]) 38 In response to the Commissions inquiry as to whether a finding of unenforceability of the 109 patent would impact the enforceability of the 353 and 405 patents, Staff argues against the application of the doctrine of infectious unenforceability. (SCN at 68-71.) Because Respondents assert that the SyncLink reference discloses the strobe signal embodiment of the asserted claims, it is not necessary to evaluate whether the doctrine of infectious unenforceability applies here. Rather, Respondents contend that SyncLink is material to the patentability of all of the asserted Barth patent claims. 39 Staff agrees that the element of intent is beyond the scope of review. (SCN at 62 n.32.)
03990.51797/4780864.12

-51-

PUBLIC VERSION

disclosed to the PTO even though it was in Rambus possession and was not subject to any confidentiality restrictions. Thus, that some information about SyncLink was disclosed during prosecution is of no consequence on the question of the materiality of RX-4270C. For these reasons, SyncLink is material to all three Barth patents, and all three should therefore be held unenforceable. VIII. ISSUE 6: DOMESTIC INDUSTRY The text of Section 337 is clear: Complainants in an ITC investigation must prove that they have made a substantial investment in licensing each of the asserted patents. See 19 U.S.C. 1337(a)(3)(c); see also RCN at 87-89. In the context of portfolio licensing, this means that a complainant must prove a strong nexus between its investment in licensing and each of the asserted patents. See Certain Multimedia Display and Navigation Devices and Systems, Inv. No. 337-TA-694, Commn Op. at 8 (Aug. 8, 2011). Without that evidence, there is no principled way for the Commission to make its required statutory finding that the claimed investment is substantial for each of the asserted patents. Rambus has not proven that its investment is substantial because it has not attributed its investment to any of the asserted patents. In Rambus view, that it has signed any licenses including the asserted patents suffices: Rambus licenses grant rights to practice the asserted patents, thereby establishing a strong nexus between them. CCN at 82. The mere fact that Rambus licenses include the asserted patents, however, does not indicate the existence of a strong nexus. The inclusion of scores of additional patents within Rambus licensespatents that may be more important than any of the asserted patentsmeans that any nexus to the asserted patents is quite weak. Rambus relies on the defined scope of Rambus license agreements to suggest that the asserted patents are important, because the scope of those license agreements is defined in such a way as to make the asserted patents significant. (CCN at 75.) But Rambus is wrong about the

03990.51797/4780864.12

-52-

PUBLIC VERSION

defined scope of its license agreements. Respondents reviewed the varying scope of Rambus license agreements in their Response to the Commissions Notice of Review. (See RCN at 9092.) These license agreements are not limited to the technologies that Rambus alleges practice the asserted patents:

(See id.) The inclusion of additional technologies that are not alleged to relate to the asserted patents makes it impossible to contend that the asserted patentsas opposed to any of Rambus other thousand-plus patents are important parts of these licenses. The Commission held in Navigation Devices that, all other things being equal, the larger the licensed portfolio, the weaker the nexus between licensing the portfolio and the asserted patents. Navigation Devices, at 11. Rambus licenses reveal why this is generally true: larger portfolio licenses, like those Rambus signed, include many patents unrelated to the patents for which Rambus must prove the existence of a domestic industry. But there is no need to make new law to find that Rambus does not meet the domestic industry requirement because its domestic industry contentions fail for a more fundamental reason Rambus not only failed to allocate its investments in licensing to the asserted patents specifically, it failed to allocate its investment even to the portfolio licenses on which it relies. The only evidence Rambus has provided of investing in licensing consist of its 2006-2010 firmwide licensing expenditures. But those investments do not relate just to the licenses it has cited as evidence of a domestic industrythe vast majority of the licenses Rambus cites pre-date the period for which it has reported investment. (See Exs. CX-39C through CX-42C, CX-44C through CX-48C, CX-50C through CX-59C, CX-579C, CX-661C, CX-662C, and CX-7628C -53-

03990.51797/4780864.12

PUBLIC VERSION

(Barth licenses); Exs. CX-4013C, CX-4326C, CX-4329C, CX-4330C, CX-4334C, CX-4338C (Dally licenses).) Nor is there any evidence indicating the extent to which Rambus subsequent firm-wide licensing investment relates to efforts to license technologies that practice the Barth or Dally patents. In other words, Rambus has failed to allocate its investment either to efforts to license the asserted patents or to efforts to license portfolios containing those patents. This is far short of the evidence in the 694 Investigation; Rambus must do more to prove the amount of its investment in licensing that is attributable to the asserted patents.40 The Staff argues that the domestic industry determination in this case is controlled by the Commissions decision in the prior 661 Investigation. (See SCN at 71-74.) But the record here differs from the record in the 661 Investigation. Here,

This is significant because Respondents here have also identified other technologies, unrelated to the Barth patents, included in purported Barth licenses; Rambus licensees may only be practicing those other, unrelated patents. Moreover, the Dally patents were not addressed in the 661 Investigation at all. Given these differences, the 661 result should not be controlling here. Rambus makes a policy argument in support of its contention that no per-patent

This failure of proof is fundamental; Rambus argument that other factors from the Navigation Devices nexus inquiry weigh in its favor is irrelevant. Even if the asserted patents were mentioned in some licensing presentations, for example, it is impossible to know what proportion of Rambus overall licensing efforts were focused on those patents. And the fact that the ALJ could identify only a handful of presentations discussing the Barth patents (see ID at 352 (cited in CCN at 82)), suggests that very little of Rambus of claimed licensing expenses went toward efforts involving the asserted patents. Even if the concept that the mere magnitude of the total company-wide licensing investment can be the basis for a finding of a domestic industry in each of the asserted patents, Rambus must at least be required to identify specific licensing activity from which it could be inferred that a substantial amount of the company-wide expenses should be attributed to the asserted patents. This Rambus failed to do; particularly for the Dally patents where the only showing of licensing activity for the claimed period was two presentations listing the asserted Dally patents among dozens of other asserted patents. (See CX7613C at .0009; CX-7612C at .0007.)
03990.51797/4780864.12

40

-54-

PUBLIC VERSION

allocation is required to prove the existence of a domestic industry, because that allocation would be inefficient and deter innovators from enforcing intellectual property. (See CCN at 75, 79-80.) This argument fails for two principal reasons. First, it is entirely unsubstantiated. Rambus never introduced evidence as to the burden that additional record keeping would entail or the extent to which an allocation requirement would deter enforcement. Second, Rambus vastly overstates the difficulty in performing a per-patent allocation. If Rambus could prove the amount it invested in securing the licenses on which it relies, or that its licensees were actually practicing the asserted patents, it could attribute efforts specifically to the asserted patents. Keeping records sufficient to link licensing investments to the asserted patents is not an undue burdenit is the minimum effort necessary prove the existence of the domestic industry in the asserted patents that is a statutory requirement to seek relief in this forum. That a potential complainant has a licensing industry does not mean that it has a licensing industry related to the asserted patents. Rambus should have to prove some more substantial connection between the asserted patents specifically and its undifferentiated firm-wide investment in licensing. IX. ISSUE 7: PATENT EXHAUSTION A. 7(a): Licensed Samsung Memory Devices Substantially Embody the Barth Patents

Rambus assertion that licensed Samsung memory devices do not substantially embody the asserted Barth claims places the syntax of the asserted claims over the reality behind Rambus infringement assertions. Patent exhaustion law does not support Rambus hypertechnical analysis. In Quanta Computer, Inc. v. LG Elecs., Inc., the Supreme Court confirmed that exhaustion is triggered by an authorized sale of a product that embodies the essential features of the patented invention if the only reasonable and intended use [is] to practice the patent. 553 U.S. 617, 631 (2007). The asserted Barth claims disclose a method for controlling a memory device and are written from the perspective of the memory controller. However, the -55-

03990.51797/4780864.12

PUBLIC VERSION

claimed method cannot be practiced without both a controller and a memory device, each is useless without the other. Therefore, a memory device is an essential part of the claimed system. Rambus infringement position relies on its assertion that JEDEC compliant DDR memory devices are always controlled, and can only be controlled, with a memory controller that practices the Barth patents. Rambus expert testified that his Barth Patent infringement analysis applies to any products that include DDR memory controllers, DDR2 memory controllers, DDR3 memory controllers, GDDR3 memory controllers, LPDDR memory controllers, and LPDDR2 memory controllers. (CX-9543C at Q171.) Rambus expert elaborated on this opinion by testifying that the JEDEC DDR memory standards require that an associated memory controller perform the claimed Barth Patent steps when interfacing with JEDEC compliant DDR memory. (See e.g. id. at Q324, 330, 334-36, 361, 366.) Thus, Rambus infringement assertions do not depend on a customized memory controller, but, instead, on the type of memory device with which the memory controller interfaces. Rambus admitted in its submission to the USPTO during the Barth patent reexaminations that the Barth patents are necessarily practiced by products that are compliant with applicable JEDEC standards. (See, e.g., RX-4243 at 42-43; RX-4257 at 32-33 (emphasis added).) In view of Rambus infringement assertions and its assertions to the USPTO, the licensed Samsung memory devices have no useful purpose other than to be incorporated into systems that practice the asserted Barth patents. See Quanta, 553 U.S. at 632 (finding an exhausting sale when the only apparent object of a component was to be incorporated into a system that practiced the asserted patents). Like in Quanta, Rambus failed to suggest any reasonable use for the licensed Samsung memory other than incorporating it into a system that practices the asserted Barth patents. Id. Thus, this case falls squarely within the facts of Quanta and authorized sales of licensed Samsung memory devices exhaust Rambus rights in the Barth patents because those -56-

03990.51797/4780864.12

PUBLIC VERSION

memory devices substantially embody the asserted Barth claims. Rambus assertions regarding testimony about nVidia memory controllers at issue during the 661 Investigation are irrelevant because they are directed at products not at issue here. However, to the extent the 661 evidence is relevant, it supports the ALJs determination that the licensed Samsung memory devices substantially embody the asserted Barth claims. The

common factor between this Investigation and the 661 investigation is that Rambus asserted memory controllers that interface with JEDEC compliant DDR memory devices infringe the Barth claims. Rambus has consistently asserted that a memory controller must operate in the claimed manner to control a JEDEC compliant memory device. Under Rambus infringement theory, a purchaser of licensed JEDEC compliant memory devices, like has no choice

but to use an infringing memory controller to interface with the licensed memory. Thus, the licensed, JEDEC compliant, Samsung memory devices substantially embody the asserted Barth claims and their authorized sales to B. 7(b): exhaust Rambus rights in the Barth patents.

Purchase of Samsung Memory Was in the United States

Rambus incorrectly focuses only on a purchase order to the exclusion of the and overlooks the relevant testimony and evidence. The governing

issued an initial purchase order after it executed the that further specified the shipment of that order as FOB Origin, . FOB is a shipping provision whereby goods are delivered at a designated location . . . where the legal title and thus risk of loss passes from seller to buyer. N. Am. Phillips Corp. v. Am. Vending Sales, Inc., 35 F.3d 1576, 1578 n.2 (Fed. Cir. 1994); U.C.C. 2-319(1)). Litecubes does not disturb the legal meaning of FOB, which in this case is Litecubes, LLC v. N. Light Prods., Inc., 523 F.3d 1353, 1358 n.1 (Fed. Cir.

03990.51797/4780864.12

-57-

PUBLIC VERSION

2008). Instead, Litecubes considered a larger set of criteria to decide the place of sale. Litecubes at 1369-71. Thus, Rambus cannot rely on Litecubes to rebut prima facie evidence of location included in the definition of FOB. Moreover, Rambus steadfastly avoids additional evidence in and the purchase orders

that is consistent with finding a sale in the United States. (RCN at 95 (the May 17, 2011 purchase order

see also TransCore, L.P. v. Elec. Transaction Consultants Corp., 563 F.3d 1271, 1278 (Fed. Cir. 2009) (evaluating the transaction as a whole). X. RAMBUS LACKS STANDING TO ASSERT THE DALLY PATENTS Respondents argued in post-hearing briefing that Rambus lacks standing to assert the Dally patents because they were automatically assigned to the University of North Carolina (UNC). (See RIB at 97-111.) Rambus and Staff contend, however, that Rambus is a bona fide purchaser of the Dally patents, and thus immune from Respondents standing challenge. (See CCN at 94-97; SCN at 80-86.) Rambus further alleges that Respondents standing challenge is barred by laches. (See CCN at 97-99.) Both arguments are mistaken. A. 8(a): Rambus Is Not a Bona Fide Purchaser

In order to be a bona fide purchaser under 35 U.S.C. 261 or otherwise, Rambus must have been without actual or inquiry notice of the UNC assignment. Bd. of Trustees of Leland Stanford Jr. Univ. v. Roche Molecular Sys., Inc., 583 F.3d 832, 843 (Fed. Cir. 2009). Rambus, however, was at least on inquiry notice of the UNC assignment. Respondents cited extensive

03990.51797/4780864.12

-58-

PUBLIC VERSION

evidence of public statements regarding the Fast Links project on which Dally and UNC professor Poulton collaborated at UNC under the UNC subcontract. (See RCN at 97-98). Both Staff and Rambus contend that these public disclosures were insufficient to notify Rambus of UNCs potential ownership interest. (See, e.g., CCN at 96-98; SCN at 82-83.) It is common knowledge that people working in such a situation will generally be required to assign their inventions to their employer or research center. See, e.g., Stanford, 563 U.S. ___, slip op. at 14 (2010) (noting that obtaining employee assignments is the common practice of federally funded contractors under the Bayh-Dole Act). Any reasonable party knowing that much of the work for the Dally patents was done by a UNC professor at UNC would have inquired further as to UNCs interests. Had Rambus made a reasonable inquiry into UNCs involvement in the Fast Links project, it would have found the UNC subcontract. Rambus had access, for example, to Dr. Poulton, who admittedly knew about the subcontract. (See CCN at 97-98 (discussing Poultons awareness).) Indeed, in the passage from FilmTec Corp. v. Allied-Signal, Inc., 939 F.2d 1568, 1574 (Fed. Cir. 1991) (cited in SCN at 84-85) (FilmTec I), cited in the Staffs brief, the court noted that the inventors awareness of the agreement containing the assignment provision could be imputed to his employer. 939 F.2d at 1574. That Dr. Poulton did not realize that the assignment agreement gave UNC a potential ownership interest is irrelevant. He is not a lawyer; Rambus should have asked him about his contractual relationship with UNC and had that contract evaluated by an attorney. If it would have done so, Rambus would have realized that UNC owned an interest in the Dally patents. Rambus cannot excuse its failure to investigate UNCs potential ownership of the Dally patents by shifting the blame to others; it did not do a diligent investigation and it failed to uncover the UNC subcontract. Even if Rambus is able to prove that it was without constructive notice of the UNC -59-

03990.51797/4780864.12

PUBLIC VERSION

subcontract, the bona fide purchaser defense cannot defeat a standing challenge. Recently, a federal district court was presented with the question of whether the bona fide purchaser defense under Section 261 could defeat a standing challenge. After reviewing the statutory history and relevant case law, the court concluded that a partys status as a bona fide purchaser is irrelevant to standing. See StemCells, Inc. v. Neuralstem, Inc., Nos. 8:06-cv-01877-AW, 8:08-cv-02664AW, 8:08-cv-01173-AW, 2012 WL 1184545, at *14-15 (D. Md. Apr. 6, 2012) (In short, the statutory history of 261 buttresses the inference that Congress did not intend for plaintiffs to be able to use 261 offensively to prove standing.). Moreover, to the extent that the Dally patents were assigned to UNC automatically pursuant to the subcontract, Dally could not have assigned them to MIT and MIT would not have title for Rambus to purchase. Section 261 is therefore unavailable. See FilmTec I, 939 F.2d at 1573; see also FilmTec Corp. v. Hydranautics, 982 F.2d 1546 (1992) (finding that statute and contract combined to effect assignment to government, depriving inventor of the right to assign to third party). B. 8(b): Laches Cannot Defeat a Standing Challenge

Rambus also argues that Respondents standing challenge is barred by the doctrine of laches. On this point, the Staff agrees with Respondents that laches cannot be used to defeat a standing challenge. (See SCN at 86-87.) The Federal Circuit has made plain that laches cannot defeat a standing claim. See Stanford, 583 F.3d at 848-49 (holding that standing challenge cannot be defeated by laches). Even if laches could apply, Rambus cannot prove its applicability here. UNC was unaware of the completion of the Dally patents and therefore could not (See RCN at 99.) Moreover, any

unreasonably delay asserting its ownership interest.

prejudice to Rambus is a result of its failure to perform a diligent investigation of UNCs involvement in the development of the Dally technologyit is not something for which Respondents should be faulted.

03990.51797/4780864.12

-60-

PUBLIC VERSION

Dated: June 1, 2012

/s/ Paul F. Brinkman Charles K. Verhoeven Sean S. Pak QUINN EMANUEL URQUHART & SULLIVAN LLP 50 California Street, 22nd Floor San Francisco, CA 94111 Thomas D. Pease QUINN EMANUEL URQUHART & SULLIVAN LLP 51 Madison Ave., 22nd Floor New York, NY 10010 Paul F. Brinkman QUINN EMANUEL URQUHART & SULLIVAN LLP 1101 Pennsylvania Avenue, 6th Floor Washington, D.C. 20004 Counsel for Respondents STMicroelectronics N.V.; STMicroelectronics Inc.; and Cisco Systems Inc. /s/ Jonathan D. Link Jonathan D. Link KILPATRICK TOWNSEND & STOCKTON LLP Suite 900 607 14th Street, NW Washington , DC 20005-2018 David E. Sipiora Daniel S. Young KILPATRICK TOWNSEND & STOCKTON LLP Suite 600 1400 Wewatta Street Denver , CO 80202 Counsel for Respondents LSI Corporation and Seagate Technology

03990.51797/4780864.12

-61-

PUBLIC VERSION

/s/ Eric C. Rusnak Eric C. Rusnak K&L GATES LLP 1601 K Street, NW Washington, DC 20006-1600 Michael J. Bettinger Steven M. Everett Curt Holbreich Elaine Y. Chow Deirdre M. Digrande K&L GATES LLP Four Embarcadero Center, Suite 1200 San Francisco, CA 94111 Counsel for Respondents STMicroelectronics, Inc. and STMicroelectronics, N.V. /s/ Andrew R. Kopsidas Andrew R. Kopsidas Ruffin Cordell FISH & RICHARDSON P.C. 1425 K Street NW, 11th Floor Washington, DC 20005 Counsel for Respondents ASUSTek Computer, Inc.; Asus Computer International, Inc.; and Hewlett- Packard Company /s/ Alexander J. Hadjis Alexander J. Hadjis MORRISON & FOERSTER LLP 2000 Pennsylvania Ave, NW Suite 6000 Washington, DC 20006 Counsel for Respondent Hitachi Global Storage Technologies, Inc.

03990.51797/4780864.12

-62-

PUBLIC VERSION

/s/ Louis S. Mastriani Louis S. Mastriani Sarah E. Hamblin Daniel F. Smith ADDUCI, MASTRIANI & SCHAUMBERG, L.L.P. 1200 Seventeenth St, NW 5th Floor Washington, DC 20036 Counsel for Respondent Garmin International Inc.

03990.51797/4780864.12

-63-

Certain Semiconductor Chips and Products Containing Same

Inv. No. 337-TA-753

CERTIFICATE OF SERVICE I, Jon Tap, hereby certify that on July 12, 2012 copies of the foregoing document were served upon the following parties as indicated: The Honorable Lisa R. Barton Acting Secretary U.S. International Trade Commission 500 E Street, SW, Room 112A Washington, DC 20436 The Honorable Theodore R. Essex Administrative Law Judge 500 E Street, SW, Room 317 Washington, DC 20436 Email: tamara.foley@usitc.gov Daniel L. Girdwood Office of Unfair Import Investigations U.S. International Trade Commission 500 E Street, SW Washington, DC 20436 Email: daniel.girdwood@usitc.gov Counsel for Complainant Rambus Inc. Christine Lehman Finnegan Henderson 901 New York Avenue, NW Washington, DC 20001 Email: ITC753-Service@finnegan.com Via First Class Mail Via Hand Delivery Via Overnight Courier Via Electronic Mail Via First Class Mail Via Hand Delivery (8 copies) Via Overnight Courier Via Electronic Mail Via E-filing Via First Class Mail Via Hand Delivery (2 copies) Via Overnight Courier Via Electronic Mail Via First Class Mail Via Hand Delivery Via Overnight Courier Via Electronic Mail

/s/ Jon Tap Jon Tap Paralegal