`
`
`
`
`
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`
`
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`
`
`INTEL CORPORATION,
`Petitioner
`
`v.
`
`QUALCOMM INCORPORATED,
`Patent Owner
`
`Patent No. 8,838,949
`
`
`
`
`Case IPR2018-01334
`
`PATENT OWNER’S SUR-REPLY ON REMAND1
`
`
`1 IPR2018-01335 and IPR2018-01336 have been consolidated with IPR2018-
`01334, and Patent Owner will file this brief only in IPR2018-01334. All citations
`are to IPR2018-01334 unless otherwise noted.
`
`
`
`
`
`IPR2018-01334
`
`TABLE OF CONTENTS
`LIST OF PATENT OWNER EXHIBITS ................................................................ ii
`I.
`INTEL’S CONSTRUCTION AND ARGUMENTS CONTRADICT
`THE INTRINSIC RECORD AND FEDERAL CIRCUIT OPINION ........... 1
`A.
`The Federal Circuit Did Not Reject The Board’s Construction .......... 1
`B.
`Intel’s Argument That The Hardware Buffer Can Be
`Implemented In System Memory Is Contrary To The Opinion ........... 2
`The Federal Circuit Found That The Hardware Buffer Provides
`Efficiency By Eliminating Extra Memory Copy Operations ............... 3
`The Federal Circuit Directed Evaluation Of Inventor Testimony ....... 5
`D.
`CLAIMS 1-9 AND 12 ARE NOT UNPATENTABLE ................................. 6
`A.
`The Board’s Finding That The ISA Is Temporary Is Dispositive ....... 6
`B.
`The ISA Formed In Internal System Memory Is The Exact
`Prior-Art Configuration Distinguished By The ’949 Patent ................ 7
`III. CONCLUSION ............................................................................................... 8
`
`C.
`
`II.
`
`
`
`-i-
`
`
`
`IPR2018-01334
`
`LIST OF PATENT OWNER EXHIBITS
`Transcript of the Deposition of Dr. Bill Lin
`Ex. 2001
`Ex. 2002 U.S. Provisional Patent Application No. 61/324,122
`Ex. 2003 Qualcomm v. Apple, Case No. 3:17-CV-1375-DMS-MDD, S.D. Cal.,
`Transcript of Jury Trial, Day 2, Volume 2-A
`Ex. 2004 Qualcomm v. Apple, Case No. 3:17-CV-1375-DMS-MDD, S.D. Cal.,
`Transcript of Jury Trial, Day 2, Volume 2-B
`Ex. 2005 Qualcomm v. Apple, Case No. 3:17-CV-1375-DMS-MDD, S.D. Cal.,
`Transcript of Jury Trial, Day 6, Volume 6-B
`Ex. 2006 Qualcomm v. Apple, Case No. 3:17-CV-1375-DMS-MDD, S.D. Cal.,
`Transcript of Jury Trial, Day 7, Volume 7-A
`Ex. 2007 Declaration of Dr. Martin Rinard
`Ex. 2008
`Transcript of Second Deposition of Dr. Bill Lin
`Ex. 2009
`Patent Owner’s Demonstratives
`
`Ex. 2010
`
`Ex. 2011 Oxford University Press, “A Dictionary of Computing” (6th ed.)
`
`Ex. 2012
`
`Lin Deposition Transcript (May 5, 2022)
`
`“Computer Architecture—A Quantitative Approach” by John L.
`Hennessy and David A. Patterson (5th ed.)
`
`
`Ex. 2013
`
`
`Ex. 2014
`
`FIFO Architecture, Functions, and Applications (Texas Instruments,
`1999)
`
`“Computer Architecture—A Quantitative Approach” by John L.
`Hennessy and David A. Patterson (4th ed.)
`
`
`Ex. 2015 Remand Declaration of Dr. Martin Rinard
`
`-ii-
`
`
`
`IPR2018-01334
`
`The Board should confirm the patentability of the challenged claims, as
`
`explained in Qualcomm’s Response on Remand (Paper 37, “PO Resp. Br.”) and here.
`
`I.
`
`INTEL’S CONSTRUCTION AND ARGUMENTS CONTRADICT
`THE INTRINSIC RECORD AND FEDERAL CIRCUIT OPINION
`The intrinsic record supports Qualcomm’s construction. PO Resp. Br. at 3-8.
`
`Intel’s construction and arguments are inconsistent with that record. Indeed, the
`
`Federal Circuit opinion directly addressed the ’949 patent specification and claims,
`
`providing significant guidance for this remand. Intel Corp. v. Qualcomm Inc., 21
`
`F.4th 801, 811 (Fed. Cir. 2021) (hereinafter, the “Opinion”). Yet, Intel’s Reply
`
`largely disregards the Federal Circuit’s opinion, or even argues contrary to it.
`
`A. The Federal Circuit Did Not Reject The Board’s Construction
`Intel misreads the Opinion in saying it “rejected” the Board’s construction.
`
`Pet. Reply at 1. The Court did not dispute the substance of the Board’s construction.
`
`Rather, it found the Board “failed to tie its construction … to the actual invention
`
`described in the specification”; “[w]hat is needed, then, is an analysis of the
`
`specification ….” Opinion at 804, 810. The Court made clear it was not “suggesting
`
`how, if at all, a proper construction will be substantively different” from the Board’s
`
`negative construction excluding temporary buffers. Id. at 810 (emphasis added).
`
`The Federal Circuit said that an analysis of the specification “seems likely to
`
`support an affirmative construction in place of the Board’s purely negative one.”
`
`Opinion at 812. To that end, Qualcomm has provided the analysis mandated by the
`
`-1-
`
`
`
`IPR2018-01334
`
`Opinion (PO Resp. Br. at 3-13), yielding a positive construction consistent with the
`
`Federal Circuit’s guidance. Intel, by contrast, disregards the Board’s previous
`
`analysis and proposes a new construction significantly different from anything
`
`considered before remand. This is wrong and unnecessary. The Court did not
`
`dispute the substance of the Board’s construction, and hence there is no reason to
`
`wipe the slate clean. The Board should bolster its analysis of the specification (as
`
`instructed by the Court) and affirm the patentability of claims 1-9 and 12 for reasons
`
`similar to those already provided in the Final Written Decision.
`
`B.
`
`Intel’s Argument That The Hardware Buffer Can Be
`Implemented In System Memory Is Contrary To The Opinion
`Intel argues that the “hardware buffer” can be implemented in system memory.
`
`See, e.g., Pet. Reply at 2. This, too, is inconsistent with the Federal Circuit’s Opinion.
`
`The Opinion states that “there must be some distinction between [the] two concepts”
`
`of “hardware buffer” and “system memory.” Opinion at 810. The Opinion also
`
`highlights “the distinctions between ‘system memory’ and ‘hardware buffer’ that are
`
`drawn both in the claim language and in the specification.” Id. at 811. Intel’s
`
`position cannot be reconciled with the Federal Circuit’s findings.
`
`Intel argues that “since the ‘hardware buffer’ must be physically separate from
`
`the claimed ‘system memory’” under its construction, there is no inconsistency with
`
`the Opinion. Pet. Reply at 5 (emphasis added). But the Court never suggested that
`
`physical separation from the claimed system memory is sufficient to constitute a
`
`-2-
`
`
`
`IPR2018-01334
`
`hardware buffer. Rather, the Court found that “Qualcomm’s ‘hardware buffer’ is …
`
`different from ‘system memory,’” and that there must be some distinction between
`
`these “concepts,” recognizing that the hardware buffer cannot be located in any
`
`system memory of the secondary processor.2 Opinion at 810-11. Intel’s reliance on
`
`dicta from the Final Written Decision (Pet. Reply at 2, citing Paper 30 at 13) is
`
`unavailing in view of the Federal Circuit’s subsequent findings. Further, the ’949
`
`patent repeatedly distinguishes conventional techniques that use a buffer formed in
`
`system memory. PO Resp. Br. at 3-13.
`
`Intel rehashes its claim-differentiation argument in support of its position that
`
`the hardware buffer can be located in system memory (Pet. Reply at 2-3), but
`
`Qualcomm has already rebutted this. PO Resp. Br. at 10-11. Claims 1 and 2 have
`
`different respective scopes, and the doctrine of claim differentiation therefore does
`
`not support Intel’s position.
`
`C. The Federal Circuit Found That The Hardware Buffer Provides
`Efficiency By Eliminating Extra Memory Copy Operations
`The Federal Circuit cited multiple portions of the ’949 patent’s specification
`
`
`2 Intel argues that its construction creates a conceptual distinction between the
`hardware buffer and the claimed system memory because the “‘hardware buffer’
`does not perform the function of a memory into which the image is loaded and
`executed.” Pet. Reply at 6. But the same distinction existed between the prior-art
`temporary buffers and the final locations in system memory from which software
`images were executed. PO Resp. Br. at 3-8. As Qualcomm showed (see id.), Intel’s
`construction of “hardware buffer” reads on the prior-art temporary buffers and is
`wrong for at least this reason.
`
`-3-
`
`
`
`IPR2018-01334
`
`that “support an understanding that use of a ‘hardware buffer’ relates to one of the
`
`key claimed advances of the invention—the elimination of ‘extra memory copy
`
`operations.’” Opinion at 811 (citing Ex. 1001 at 5:31-35, 7:16, 7:27-30, 9:42-46).
`
`Intel ignores the Court’s finding and the specification’s disclosure. It argues
`
`that extra memory copy operations are not eliminated in the invention of the ’949
`
`patent because “[t]here is always an intermediate copying step as the executable
`
`image moves from the primary processor to the hardware buffer, and then to the
`
`claimed ‘system memory’….” Pet. Reply at 3. Even accepting arguendo Intel’s
`
`view that copying is involved with the hardware buffer, that is immaterial. The
`
`copying that the ’949 patent seeks to avoid is copying in system memory. PO Resp.
`
`Br. at 3-8. The ’949 patent describes that these copy operations in system memory
`
`result in reduced efficiency and increased boot times (see, e.g., Ex. 1001 at 7:16-30),
`
`whereas the invention uses the hardware buffer to eliminate these copy operations
`
`in system memory and provide “a more efficient direct loading process” (see, e.g.,
`
`id. at 9:43-50). Ex. 2015 (Rinard Remand Decl.) at ¶¶ 23-34.
`
`Notably, Intel does not challenge (much less rebut) Qualcomm’s showings
`
`about the differences between system memory and the hardware buffer of the ’949
`
`patent. Dr. Rinard provided detailed testimony, supported by documentary evidence
`
`(id. at ¶¶ 6-15), that system memory is general-purpose and is not specifically
`
`configured to buffer data in a multi-processor system. Id. at ¶¶ 25-26. He further
`
`-4-
`
`
`
`IPR2018-01334
`
`explained that the ’949 patent’s hardware buffer is permanently assigned within the
`
`structure of the hardware to perform the single task of loading data segments directly
`
`to system memory. Id. at ¶¶ 31-34. All of this testimony stands unrebutted.
`
`Rather than address Qualcomm’s showings head-on, Intel argues that “an
`
`internal RAM (even if in system memory)” would buffer data “as quickly and
`
`efficiently as … a permanent [hardware] buffer.” Pet. Reply at 4, 7. Therefore,
`
`according to Intel, “the term ‘hardware buffer’ should not be construed to preclude
`
`the use of such an internal RAM as a ‘hardware buffer,’ regardless of whether it
`
`might be characterized as part of some system memory.” Id. at 4. But this is wrong
`
`because the ’949 patent explicitly distinguishes prior-art approaches that use internal
`
`RAM of system memory for buffering data in a multi-processor system. Ex. 1001
`
`at 2:31-34 (describing prior-art approach in which “[t]he temporary buffer would be
`
`some place in system memory, such as in internal random-access-memory
`
`(RAM) ….”), 4:46-47 (“[T]he direct scatter load technique [of the invention] avoids
`
`use of a temporary buffer.”). These portions make clear that the hardware buffer
`
`cannot be located in internal RAM of system memory, and the alleged efficiency of
`
`internal RAM therefore cannot transform it into a hardware buffer.
`
`D. The Federal Circuit Directed Evaluation Of Inventor Testimony
`Intel criticizes Qualcomm’s citation to inventor testimony as “self-serving”
`
`and “irrelevant.” Pet. Reply at 5, Ex. 1027 at ¶ 16. But the Federal Circuit directed
`
`-5-
`
`
`
`IPR2018-01334
`
`the Board to consider this testimony—it “would benefit from attention on remand.”
`
`Opinion at 812. The inventor testimony is relevant and evidences Qualcomm’s
`
`desire to develop an efficient loading process that avoids “copying things around
`
`memory” (Ex. 2003 at 216:16-21), consistent with the description of the ’949 patent.
`
`II. CLAIMS 1-9 AND 12 ARE NOT UNPATENTABLE
`A. The Board’s Finding That The ISA Is Temporary Is Dispositive
`Under the correct construction of “hardware buffer” (i.e., “a permanent,
`
`dedicated buffer that is distinct from system memory”), Intel’s patentability
`
`challenges to claims 1-9 and 12 fail because the Board has already determined that
`
`the intermediate storage area (ISA) of Svensson/Bauer is temporary, i.e., not
`
`permanent. As Qualcomm explained (PO Resp. Br. at 3, 16), the Federal Circuit did
`
`not disturb this finding that the ISA is temporary, and it should stand on remand.
`
`Intel suggests this issue remains open, arguing that the Federal Circuit vacated
`
`the Final Written Decision as to claims 1-9 and 12 and that Intel’s appeal challenged
`
`the ISA as a temporary buffer. Pet. Reply at 8 (internal quotation omitted). But the
`
`Federal Circuit vacated and remanded “for reconsideration of claim construction”
`
`(Opinion at 812), not for the Board to reconsider its (correct) finding that
`
`Svensson/Bauer’s ISA is a temporary buffer. As the Board explained, “Svensson
`
`discloses that the intermediate storage area is reserved at runtime,” and “[t]hus, we
`
`find that the intermediate storage area of Bauer and Svensson is a temporary buffer.”
`
`-6-
`
`
`
`IPR2018-01334
`
`FWD at 56 (citing Ex. 1010 (Svensson) at 5:21-29; see also Paper 29 (Hearing Tr.)
`
`at 19:19-25 (Intel agreeing that the ISA is reserved at runtime). The Board’s analysis
`
`and findings are correct and consistent with Dr. Rinard’s testimony that the ISA is
`
`temporary because “[i]t exists only after it is allocated by software at run time, and
`
`therefore, it is not a permanent, fixed buffer.” Ex. 2015 at ¶ 65.
`
`Intel raises its deallocation argument again (Pet. Reply at 7), but Qualcomm
`
`has already rebutted this—the ISA of Svensson/Bauer is no different than the prior-
`
`art temporary buffers described in the ’949 patent, which likewise makes no mention
`
`of deallocation (PO Resp. Br. at 16-17)—and the Reply raises no new arguments on
`
`this point. In any event, just because Svensson does not explicitly “describe[]”
`
`deallocation of the ISA (Pet. Reply at 7), it does not follow that the ISA exists forever
`
`(i.e., permanently). The ISA is allocated in the volatile DSP SARAM &
`
`DARAM 108/208 and therefore will cease to exist at least when the multi-processor
`
`system is powered down, if not sooner. It is not a permanent buffer.
`
`B.
`
`The ISA Formed In Internal System Memory Is The Exact Prior-
`Art Configuration Distinguished By The ’949 Patent
`Intel argues that because the ISA of Svensson/Bauer is formed in internal DSP
`
`SARAM & DARAM 108/208, the ISA necessarily operates “as quickly and
`
`efficiently as copying from a permanent buffer.” Pet. Reply at 7. This argument
`
`amounts to mere speculation. Specifically, although Intel argues that internal
`
`memory implemented as SRAM is “much faster” than DRAM (Ex. 1027 at ¶¶ 19,
`
`-7-
`
`
`
`IPR2018-01334
`
`23), neither Svensson nor Bauer describes the DSP SARAM & DARAM 108/208 as
`
`being SRAM. Likewise, Intel’s argument that cache memory is “typically
`
`implemented using SRAM” and “function[s] at or near the speed of the CPU” (id.
`
`at ¶¶ 22-24) is irrelevant to the Svensson/Bauer ground because these references do
`
`not describe the memory 108/208 or its ISA as a cache. Any advantages of SRAM
`
`or cache memory cannot be imputed to the ISA.
`
`In any event, the alleged efficiency of the ISA is beside the point. The ISA is
`
`indistinguishable from the prior-art temporary buffers formed in system memory.
`
`Compare Ex. 1001 (’949 patent) at 2:23-34 (“[O]ne way of performing such loading
`
`is to allocate a temporary buffer …. From the temporary buffer, … the payload
`
`would get copied over to the final destination. The temporary buffer would be some
`
`place in system memory, such as in internal random-access-memory (RAM) ….”)
`
`to Ex. 1010 (Svensson) at 5:21-29 (“The idle process reserves a block of memory
`
`in … ‘internal’ memory …. [T]his reserved block of memory is used for intermediate
`
`storage of information (code and/or data) to be transferred to the slave-private
`
`memory.”); see also PO Resp. Br. at 15-19. If the term “hardware buffer” could
`
`encompass the ISA of Svensson/Bauer, then the claims would read on the prior-art
`
`techniques distinguished throughout the ’949 patent. That cannot be correct.
`
`III. CONCLUSION
`
`The Board should hold that claims 1-9, 12, 16, and 17 are not unpatentable.
`
`-8-
`
`
`
`IPR2018-01334
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Date: July 13, 2022
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Respectfully submitted,
`
`
`
`By: /David B. Cochran/
`Joseph M. Sauer, Reg. No. 47,919
`David B. Cochran, Reg. No. 39,142
`David M. Maiorana, Reg. No. 41,449
`Matthew W. Johnson, Reg. No. 59,108
`Joshua R. Nightingale, Reg. No. 67,865
`JONES DAY
`North Point, 901 Lakeside Avenue
`Cleveland, OH 44114
`(216) 586-3939
`
`-9-
`
`
`
`IPR2018-01334
`
`CERTIFICATE OF SERVICE
`
`The undersigned hereby certifies that the foregoing Patent Owner Sur-Reply
`
`on Remand was served via email on the date below, upon the following:
`
`David L. Cavanaugh
`David.Cavanaugh@wilmerhale.com
`
`Thomas E. Anderson
`Tom.Anderson@wilmerhale.com
`
`Joseph H. Haag
`Joseph.Haag@wilmerhale.com
`
`
`Dated: July 13, 2022
`
`
`/Joshua R. Nightingale/
`Counsel for Patent Owner
`
`
`
`
`
`
`
`
`