`
`
`
`
`
`
`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 RESPONSE BRIEF 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
`
`B.
`
`C.
`D.
`
`TABLE OF CONTENTS
`TABLE OF AUTHORITIES ................................................................................... ii
`LIST OF PATENT OWNER EXHIBITS ............................................................... iii
`I.
`INTRODUCTION .......................................................................................... 1
`II.
`CLAIM CONSTRUCTION—“HARDWARE BUFFER” ............................ 3
`A.
`The ’949 Patent Describes Conventional Loading Processes
`That Use Temporary Buffers In System Memory ............................... 3
`The ’949 Invention’s Use Of A Dedicated, Permanent
`Hardware Buffer Enables More Efficient Loading .............................. 6
`Inventor Testimony Supports Qualcomm’s Construction ................... 8
`Intel’s Arguments Contrary To Qualcomm’s Construction Are
`Wrong ................................................................................................... 9
`Intel’s Proposed Construction Is Overly Broad And Does Not
`Capture The Scope Of The Actual Invention ..................................... 14
`III. CLAIMS 1-9 AND 12 ARE NOT UNPATENTABLE ............................... 15
`A.
`The ISA Is Not A Permanent Buffer .................................................. 16
`B.
`The ISA Is Not A Dedicated Buffer Distinct From System
`Memory .............................................................................................. 17
`IV. CLAIMS 16 AND 17 ................................................................................... 19
`V.
`CONCLUSION ............................................................................................. 20
`
`E.
`
`
`
`-i-
`
`
`
`TABLE OF AUTHORITIES
`
`IPR2018-01334
`
`Page
`
`
`
`CASES
`
`3M Innovative Props. Co. v. Avery Dennison Corp.,
`350 F.3d 1365 (Fed. Cir. 2003) .......................................................................... 15
`
`Fantasy Sports Props., Inc. v. Sportsline.com, Inc.,
`287 F.3d 1108 (Fed. Cir. 2002) .......................................................................... 10
`
`Intel Corp. v. Qualcomm Inc.,
`21 F.4th 801 (Fed. Cir. 2021) ......................................................................passim
`
`Laitram Corp. v. Rexnord, Inc.,
`939 F.2d 1533 (Fed. Cir. 1991) .......................................................................... 10
`
`Salazar v. Procter & Gamble Co.,
`414 F.3d 1342 (Fed. Cir. 2005) .......................................................................... 15
`
`
`
`-ii-
`
`
`
`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
`
`-iii-
`
`
`
`IPR2018-01334
`
`I.
`
`INTRODUCTION
`The primary issue before the Board on remand is the meaning of the claim
`
`term “hardware buffer.” Claims 1-9 of the ’949 patent are directed to a multi-
`
`processor system including “system memory and a hardware buffer for receiving an
`
`image header and at least one data segment of an executable software image.” The
`
`claimed multi-processor system also includes a scatter loader controller configured
`
`“to scatter load each received data segment … directly from the hardware buffer to
`
`the system memory.” Dependent claim 2 further recites that “the scatter loader
`
`controller is configured to load the executable software image directly from the
`
`hardware buffer to the system memory of the secondary processor without copying
`
`data between system memory locations on the secondary processor.”
`
`Analyzing this claim language, the Federal Circuit determined that “the
`
`meaning of ‘hardware buffer’ relates to the ability to move the software image
`
`‘directly’ to the second processor’s system memory and to avoid ‘copying data
`
`between system memory locations.’” Intel Corp. v. Qualcomm Inc., 21 F.4th 801,
`
`810 (Fed. Cir. 2021) (hereinafter, the “Opinion”). The court further found that
`
`“because claim 1 requires both a ‘system memory’ and a ‘hardware buffer,’ there
`
`must be some distinction between those two concepts.” Id.
`
`The Federal Circuit determined, however, that these “conclusions from the
`
`claim language advance the claim-construction inquiry only so far.” Id. “What is
`
`-1-
`
`
`
`IPR2018-01334
`
`needed, then is an analysis of the specification to arrive at an understanding of what
`
`it teaches about what a ‘hardware buffer’ is.” Id.; see id. at 811 (“What is needed in
`
`this case is a more substance-focused analysis than is yet present . . . .”). In this brief,
`
`Qualcomm provides the “substance-focused analysis” that the Federal Circuit found
`
`missing from the Board’s decision. In view of the claim language and the analysis
`
`herein, the term “hardware buffer” should be construed as a permanent, dedicated
`
`buffer that is distinct from system memory. This construction is consistent with the
`
`Federal Circuit’s decision and captures what the inventors actually invented and
`
`intended to envelop with the claim. See id. at 809.
`
`Intel’s proposed construction, by contrast, is divorced from the invention
`
`described and claimed in the ’949 patent. Intel now asserts that the claimed
`
`“hardware buffer” can be located in system memory (Intel Br. at 6), but the Federal
`
`Circuit has already rejected that argument. Opinion at 810-11. Moreover, a
`
`hardware buffer formed in system memory would eviscerate the invention described
`
`in the ’949 patent. As recognized by the Federal Circuit, the “use of a ‘hardware
`
`buffer’ relates to one of the key claimed advances of the invention—the elimination
`
`of ‘extra memory copy operations’” in system memory. Opinion at 811. Under
`
`Intel’s construction, extra memory copy operations in system memory are not
`
`eliminated, and the advantages of the ’949 invention are not realized. Intel’s
`
`construction is wrong for these reasons and those detailed below.
`
`-2-
`
`
`
`IPR2018-01334
`
`Under Qualcomm’s correct construction of “hardware buffer,” claims 1-9
`
`and 12 are not unpatentable over Bauer, Svensson, and Kim. Intel argues that the
`
`intermediate storage area (“ISA”) of Bauer and Svensson discloses the claimed
`
`“hardware buffer” (Intel Br. at 13), but the Board already found that the ISA in those
`
`references is a temporary buffer—i.e., not permanent. The Federal Circuit did not
`
`disturb this finding, and it should stand on remand. The ISA of Bauer and Svensson
`
`also fails to satisfy Qualcomm’s construction because it is formed in general-purpose
`
`system memory and therefore is not a dedicated buffer distinct from system memory.
`
`As to claims 16 and 17, the Federal Circuit held that the Board must make its
`
`own determination about whether there is sufficient structure in the specification to
`
`support the means-plus-function (MPF) limitations of these claims. Opinion at 813-
`
`14. Qualcomm accordingly rests on the prior briefing for this issue. If the Board
`
`determines that the specification lacks sufficient structure, then the Board cannot
`
`resolve the prior-art challenges to these claims, as detailed below.
`
`II. CLAIM CONSTRUCTION—“HARDWARE BUFFER”
`The term “hardware buffer” should be construed as “a permanent, dedicated
`
`buffer that is distinct from system memory.” Ex. 2015 at ¶¶16-61.
`
`A. The ’949 Patent Describes Conventional Loading Processes That
`Use Temporary Buffers In System Memory
`The Background section of the ’949 patent describes prior-art systems in
`
`which a software image is loaded onto a target, secondary processor from a primary
`
`-3-
`
`
`
`IPR2018-01334
`
`processor using “an intermediate step where the binary multi-segmented image is
`
`transferred into the system memory and then later transferred into target locations
`
`by the boot loader.” Ex. 1001 at 2:17-22; Ex. 2015 at ¶¶23-24. This loading is
`
`performed by “allocat[ing] a temporary buffer into which each packet is received.”
`
`Ex. 1001 at 2:25-26. The temporary buffer (also referred to as an “intermediate
`
`buffer,” see id. at 2:35-54) is “some place in system memory, such as in internal
`
`random-access-memory (RAM) or double data rate (DDR) memory, for example.”
`
`Id. at 2:31-34.
`
`The ’949 patent explains that the prior-art use of system memory for buffering
`
`is inefficient because it requires “extra memory copy operations.” Ex. 1001 at 7:16-
`
`30. Specifically, the approaches described in the Background section of the ’949
`
`patent involve the intermediate step of copying image data from the temporary
`
`buffer—which is formed in one portion of system memory, see id. at 2:31-34—to
`
`its final location in system memory. Id. at 2:17-22, 2:29-31, 2:35-41, 7:20-26;
`
`Ex. 2015 at ¶24. The ’949 patent states that the extra memory copy operations in
`
`system memory result in reduced efficiency and increased time required to boot a
`
`secondary processor in a multi-processor system. Ex. 1001 at 7:27-30.
`
`As Dr. Rinard explains (Ex. 2015 at ¶¶25-26), the prior-art systems’ use of
`
`system memory is inefficient also because it is general-purpose memory that is not
`
`specifically configured for the task of buffering data in a multi-processor system.
`
`-4-
`
`
`
`IPR2018-01334
`
`System memory is general-purpose working memory of a processor, usable for a
`
`variety of different tasks. Id. at ¶25. Under software control, the processor allocates
`
`different regions of system memory for the various data and code storage tasks
`
`required to execute the software. Id. The temporary buffer described in the
`
`Background section of the ’949 patent is an example of a buffer that is allocated by
`
`software in system memory, see Ex. 1001 at 2:23-28, and this buffer is “temporary”
`
`because it only exists (conceptually) after it is allocated by the software. Ex. 2015
`
`at ¶25.
`
`Different software running on the processor can apply different configurations
`
`to the system memory, and the system memory is therefore capable of supporting a
`
`wide variety of different access patterns. Id. at ¶26. The ability to support different
`
`access patterns provides flexibility, but the flexibility comes at the cost of speed and
`
`efficiency. Id.; see also id. at ¶¶6-15. Because system memory—including the
`
`system memory allocated for use as temporary buffers in the prior-art approaches,
`
`see Ex. 1001 at 2:23-34—is able to support a wide range of software memory access
`
`patterns, using system memory for buffering is less efficient than using dedicated
`
`hardware components for this purpose. Ex. 2015 at ¶26. The hardware buffer of
`
`the ’949 patent is an example of a dedicated hardware component, distinct from
`
`system memory, that provides “improv[ed] performance” over the conventional
`
`approaches, as explained below. Id.; Ex. 1001 at 7:27-30.
`
`-5-
`
`
`
`IPR2018-01334
`
`B.
`
`The ’949 Invention’s Use Of A Dedicated, Permanent Hardware
`Buffer Enables More Efficient Loading
`The invention of the ’949 patent uses a hardware buffer to provide “a direct
`
`scatter load technique” that eliminates “the intermediate step of buffering required
`
`in traditional loading processes,” to achieve increased efficiency. Ex. 1001 at 4:43-
`
`47, 7:20-26; Ex. 2015 at ¶¶27-34. The ’949 patent refers to the direct scatter load
`
`technique as a “Zero Copy Transport flow” (Ex. 1001 at 7:16), an example of which
`
`is illustrated in Figure 3. In Figure 3, a scatter load controller 304 transfers image
`
`segments “from the hardware buffer of the hardware transport mechanism 309
`
`directly into their respective target locations in the … system memory 305.” Id.
`
`at 9:21-27; see also Ex. 2015 at ¶¶27-28.
`
`In transferring data directly from the hardware buffer to the target locations in
`
`system memory, the invention of the ’949 patent eliminates the system memory to
`
`system memory copying operations of the prior art. Ex. 2015 at ¶29. Specifically,
`
`the ’949 patent states that in performing the direct transfer using the hardware buffer,
`
`“no extra memory copy operations occur in the secondary processor.” Ex. 1001
`
`at 9:42-43. The patent contrasts the invention with the “conventional techniques
`
`employing a temporary buffer,” stating that the conventional techniques “are
`
`bypassed in favor of a more efficient direct loading process.” Id. at 9:43-46. The
`
`direct loading process “does not require the intermediate buffer operations
`
`-6-
`
`
`
`IPR2018-01334
`
`traditionally required for loading a software image from a primary processor to a
`
`secondary processor.” Id. at 9:46-50.
`
`In avoiding the extra memory copy operations of the prior art, the invention
`
`provides a more efficient loading process. Ex. 2015 at ¶30. The ’949 patent states,
`
`for instance, that the direct scatter load technique of the invention “reduce[s] the
`
`time it takes to boot secondary processors in a multi-processor system where
`
`secondary processor images are transferred from the primary processor. This
`
`reduction is achieved by avoiding extra memory copy operations ….” Ex. 1001
`
`at 11:17-24; see also id. at 2:23-55, 4:46-47, 5:31-35, 7:16-30, 9:9-11.
`
`The hardware buffer of the ’949 patent also enables a more efficient loading
`
`process because it is permanent and dedicated to the single task of loading data
`
`segments directly to system memory. Ex. 2015 at ¶¶32-33. As explained above, the
`
`conventional techniques distinguished in the ’949 patent use a temporary buffer that
`
`is allocated by software in general-purpose system memory. The prior-art buffer is
`
`temporary—i.e., not permanent—because it exists only after software allocates it,
`
`and its general-purpose nature makes it flexible but relatively inefficient. Id. at ¶32.
`
`By contrast, in the invention of the ’949 patent, the hardware buffer is permanently
`
`assigned within the structure of the hardware to perform the single, specialized task
`
`for which it was designed, i.e., the direct transfer of data from a primary processor
`
`to the system memory of a secondary processor. Id.; see also id. at ¶¶6-15. The
`
`-7-
`
`
`
`IPR2018-01334
`
`permanent and dedicated nature of the hardware buffer—along with the elimination
`
`of “extra memory copy operations”—enables the efficient direct scatter loading
`
`technique described and claimed in the ’949 patent. Id. at ¶¶32-33.
`
`For these reasons, the term “hardware buffer” should be construed as a
`
`permanent, dedicated buffer that is distinct from system memory. Id. at ¶34.
`
`C.
`Inventor Testimony Supports Qualcomm’s Construction
`Steve Haehnichen, a named inventor on the ’949 patent, provided testimony
`
`relevant to the meaning of the term “hardware buffer.” See generally Exs. 2003-
`
`2004; see also Ex. 2015 at ¶¶35-40 (Dr. Rinard analyzing Mr. Haehnichen’s
`
`testimony). Qualcomm agrees with the Federal Circuit that this testimony “would
`
`benefit from attention on remand,” Opinion at 812, because it provides evidence of
`
`what the inventors actually invented and intended to envelop with the claim.
`
`Mr. Haehnichen testified that Qualcomm was seeking to design a modem
`
`processor with a boot time of less than one second. Ex. 2003 at 213:22-214:2 (“Boot
`
`time should be 1 second”), 214:1-12. To meet the one-second boot time requirement,
`
`Qualcomm developed a new design (referred to as “Sahara”) that became the basis
`
`for the ’949 patent. Id. at 215:20-216:15. Mr. Haehnichen testified that “[t]he single
`
`biggest thing we did to speed it up was to avoid copying things around memory.”
`
`Id. at 216:16-21; see also id. at 216:22-217:2. Mr. Haehnichen also contrasted the
`
`Sahara design with Qualcomm’s previous “Gobi” design: “So if you look at Gobi, it
`
`-8-
`
`
`
`IPR2018-01334
`
`reads things into memory, figures out what they are, where they need to go, and it
`
`copies them there; whereas in Sahara, we would receive the data directly from USB
`
`and the only place it would go is right where it needs to be. So we compared to like
`
`a nonstop flight. Instead of a flight with a stop in the middle and a long layover, you
`
`just put it right where it needs to be at the end.” Ex. 2004 at 221:25-222:10.
`
`Mr. Haehnichen’s testimony reflects the fact that the permanent hardware
`
`buffer of the ’949 patent enables a more efficient, direct transfer that outperforms
`
`less efficient techniques that involve “copying things around memory.” Ex. 2003
`
`at 216:16-21; Ex. 2015 at ¶¶39-40. Further, the permanent, dedicated nature of the
`
`hardware buffer enables it to be integrated into an efficient direct data transfer
`
`mechanism, such as the USB controller described in the ’949 patent. Ex. 2003
`
`at 216:22-217:2 (Mr. Haehnichen testifying that “USB speed is fast”).
`
`D.
`
`Intel’s Arguments Contrary To Qualcomm’s Construction Are
`Wrong
`Intel’s brief includes several arguments relevant to Qualcomm’s construction.
`
`These arguments are wrong for the reasons detailed below. Ex. 2015 at ¶¶41-56.
`
`Intel asserts that nothing in the ’949 patent precludes the hardware buffer from
`
`being located in system memory (Intel Br. at 6), but the Federal Circuit has already
`
`determined otherwise. The court recognized that “claim 1 requires both a ‘system
`
`memory’ and a ‘hardware buffer,’” and therefore, “there must be some distinction
`
`between those two concepts.” Opinion at 810; see also id. at 811. Moreover, Intel’s
`
`-9-
`
`
`
`IPR2018-01334
`
`argument is contrary to the “actual invention described in the specification” (id.
`
`at 804), which eliminates copy operations in system memory to provide increased
`
`efficiency and reduced boot times. See Section II.B above. If the hardware buffer
`
`could be located in system memory (as Intel argues), then these advantages are not
`
`realized, and the claims would read on the prior-art techniques distinguished in
`
`the ’949 patent. Ex. 2015 at ¶¶42-43. That cannot be correct.
`
`Intel makes a claim-differentiation argument, asserting that dependent claim 2
`
`“confirms that neither claim 1 nor its ‘hardware buffer’ limitation precludes the
`
`possibility of copying data between system memory locations.” Intel Br. at 7. For
`
`several reasons, this is incorrect. As an initial matter, claim differentiation is a guide,
`
`not a rigid rule, Laitram Corp. v. Rexnord, Inc., 939 F.2d 1533, 1538 (Fed. Cir. 1991),
`
`and claim differentiation cannot broaden claims beyond their correct scope. Fantasy
`
`Sports Props., Inc. v. Sportsline.com, Inc., 287 F.3d 1108, 1115-16 (Fed. Cir. 2002).
`
`Intel’s purported claim-differentiation argument fails because it would improperly
`
`broaden claims 1 and 3-9 beyond their correct scope. Ex. 2015 at ¶44. As detailed
`
`above and recognized by the Federal Circuit, the “use of a ‘hardware buffer’ relates
`
`to one of the key claimed advances of the invention—the elimination of ‘extra
`
`memory copy operations’” in system memory. Opinion at 811. Intel’s claim-
`
`differentiation position—which, if adopted, would yield a claim that permits
`
`“copying data between system memory locations,” see Intel Br. at 7—would
`
`-10-
`
`
`
`IPR2018-01334
`
`improperly broaden these claims beyond the scope of the invention described in the
`
`specification. Ex. 2015 at ¶44.
`
`Further, Intel’s purported claim-differentiation argument fails because claim
`
`differentiation exists under Qualcomm’s construction of “hardware buffer.” Id.
`
`at ¶45. With Qualcomm’s construction of “hardware buffer” inserted, claim 1
`
`requires a scatter loader controller configured to “scatter load each received data
`
`segment based at least in part on the loaded image header, directly from the
`
`[permanent, dedicated buffer that is distinct from system memory] to the system
`
`memory.” Claim 2 further limits the scatter loader controller element by adding
`
`limitations on how it “directly” loads into system memory—specifically, expressly
`
`excluding “copying data between system memory locations on the secondary
`
`processor.” Thus, claims 1 and 2 have different respective scopes, and there is no
`
`claim-differentiation issue.
`
`Intel disagrees that the hardware buffer must be permanent, arguing that
`
`the ’949 patent does not “distinguish … the use of temporary buffers per se.” Intel
`
`Br. at 9. Rather, according to Intel, the specification “distinguishes systems in which
`
`the entire executable image is copied into a temporary buffer.” Id. (boldface and
`
`italics by Intel). That is incorrect. The ’949 patent also distinguishes prior-art
`
`approaches in which a temporary buffer receives “part of the image data.” Ex. 1001
`
`at 2:23-55; Ex. 2015 at ¶¶46-47. Thus, the techniques distinguished by the ’949
`
`-11-
`
`
`
`IPR2018-01334
`
`patent are not limited to those that involve copying an entire executable image into
`
`a temporary buffer.
`
`Intel also argues that the specification distinguishes “between the buffer
`
`receiving a header separately from data segments (as claimed) and receiving a
`
`packet that includes both a header and data segments, and not that the buffer is
`
`temporary.” Intel. Br. at 10 (boldface and italics by Intel). This, too, is wrong.
`
`Ex. 2015 at ¶¶48-53. First, the ’949 patent states, without qualification, that “the
`
`direct scatter load technique avoids use of a temporary buffer.” Ex. 1001 at 4:46-47.
`
`Second, the patent describes a modem processor 110 that “stores the modem
`
`executable image 132 directly … to the final destination without copying the data
`
`into a temporary buffer in the modem processor RAM 112.” Id. at 5:31-35. These
`
`statements distinguish the use of a temporary buffer in system memory and do not
`
`limit the distinction only to temporary buffers that receive packets with both a header
`
`and data segments. Ex. 2015 at ¶53.
`
`Intel’s attempts to avoid these two statements fail. Id. at ¶¶54-56. Intel argues
`
`that the first statement (at Ex. 1001 at 4:46-47) must be “read in light of the
`
`surrounding context,” and that the surrounding context shows that the ’949 patent
`
`“distinguishes receiving a packet that includes both a header and data segments, not
`
`the use of temporary buffers for all purposes.” Intel Br. at 11. But the statement at
`
`column 4, lines 46-47 is categorical and does not link or limit avoidance of the
`
`-12-
`
`
`
`IPR2018-01334
`
`temporary buffer to systems that use packets containing headers and data. Ex. 2015
`
`at ¶54. The statement is also consistent with other portions of the ’949 patent (see,
`
`e.g., Ex. 1001 at 5:31-35, 7:16-30, 11:11-24) describing the elimination of temporary
`
`buffers in system memory without reference to the packet/header distinction that
`
`Intel tries to make. Further, the relevant “context” for the statement at column 4,
`
`lines 46-47 is the patent’s discussion of the “Zero Copy Transport flow” in column 7,
`
`which describes “alleviat[ing] the intermediate step of buffering required in
`
`traditional loading processes” without mentioning packets or headers. Ex. 1001
`
`at 7:16-30; Ex. 2015 at ¶54.
`
`Regarding the second statement (at Ex. 1001 at 5:31-35), Intel argues that “the
`
`specification describes transferring the executable image into the final location in
`
`the system memory (i.e., RAM 112) without copying the data into a temporary
`
`buffer in the same system memory (i.e., RAM 112).” Intel Br. at 11 (boldface and
`
`italics by Intel). Thus, Intel argues that the hardware buffer cannot be located in the
`
`RAM 112 but could be located in some other system memory of the modem
`
`processor 110. Id. But as Dr. Rinard explains (Ex. 2015 at ¶¶55-56), the patent
`
`precludes the hardware buffer from being in the only system memory of the
`
`processor 110 (i.e., RAM 112), and it would be illogical for the patent to address
`
`(nonexistent) system memories separate from the RAM 112 that are not described
`
`in the patent.
`
`-13-
`
`
`
`IPR2018-01334
`
`E.
`
`Intel’s Proposed Construction Is Overly Broad And Does Not
`Capture The Scope Of The Actual Invention
`Intel argues that the claimed “hardware buffer” is “memory that is physically
`
`separate from the memory into which the software image is loaded for execution.”
`
`Intel Br. at 5. Intel argues that its construction “does not foreclose the possibility of
`
`implementing a [hardware] buffer in some other system memory.” Id. at 6.
`
`Intel’s construction is wrong for several reasons. Ex. 2015 at ¶¶57-61. First,
`
`Intel’s construction encompasses a buffer formed in system memory, which is the
`
`exact configuration described in the Background section and distinguished
`
`throughout the ’949 patent. Id. at ¶59; Sections II.A-B above.
`
`Second, Intel’s construction is contrary to the Federal Circuit’s finding that
`
`the hardware buffer is conceptually distinct from system memory. Opinion at 810.
`
`The court did not state that the hardware buffer and system memory must merely be
`
`physically distinct; rather, the court stated that “there must be some distinction
`
`between those two concepts.” Id. Locating the hardware buffer in system memory
`
`(as is permissible under Intel’s construction) would eliminate the conceptual
`
`distinction between the two components and fail to achieve the “Zero Copy
`
`Transport Flow” that the inventors intended to envelop. Ex. 2015 at ¶60.
`
`Third, Intel’s construction is contrary to positions it took previously. Intel’s
`
`declarant Dr. Lin has acknowledged that “[t]he ’949 patent makes a distinction
`
`between prior art systems that used a ‘temporary buffer’ that was part of the system
`
`-14-
`
`
`
`IPR2018-01334
`
`memory … and the alleged invention that uses a ‘hardware buffer’ separate from the
`
`system memory.” Ex. 1002 at ¶111. Intel now abandons that position, arguing that
`
`the hardware buffer can be in system memory. Intel Br. at 6. This shows that Intel’s
`
`shifting positions are nothing more than litigation-induced maneuvering.
`
`Finally, Intel wrongly argues that the prosecution history supports its
`
`proposed construction. Intel argues that “the Examiner expressly identified
`
`Svensson’s ISA as a ‘hardware buffer,’” and Applicant did not “argu[e] that
`
`Svensson lacked a ‘hardware buffer.’” Intel Br. at 11-12. But unilateral statements
`
`by an Examiner cannot redefine claim terms, Salazar v. Procter & Gamble Co., 414
`
`F.3d 1342, 1344-48 (Fed. Cir. 2005), and “[a]n applicant’s silence in response to an
`
`examiner’s characterization of a claim does not reflect the applicant’s clear and
`
`unmistakable acquiescence to that characterization if the claim is eventually allowed
`
`on grounds unrelated to the examiner’s unrebutted characterization.” 3M Innovative
`
`Props. Co. v. Avery Dennison Corp., 350 F.3d 1365, 1373-74 (Fed. Cir. 2003).
`
`The ’949 claims were allowed for reasons unrelated to the Examiner’s mistaken
`
`belief that Svensson’s ISA is a hardware buffer, and Applicant was therefore not
`
`acquiescing to the Examiner’s position. See Paper 16 (“PO Resp.”) at 8.
`
`III. CLAIMS 1-9 AND 12 ARE NOT UNPATENTABLE
`Under Qualcomm’s correct construction of “hardware buffer,” claims 1-9
`
`and 12 are not unpatentable over Bauer, Svensson, and Kim. Intel argues that the
`
`-15-
`
`
`
`IPR2018-01334
`
`ISA of Bauer and Svensson discloses the claimed “hardware buffer” (Intel Br. at 13),
`
`but this is wrong for the reasons detailed below. Ex. 2015 at ¶¶62-70.
`
`A. The ISA Is Not A Permanent Buffer
`Qualcomm previously showed that Bauer/Svensson’s ISA is a temporary
`
`buffer reserved in system memory at run time and used to hold data that is
`
`subsequently transferred to a final destination in system memory. PO Resp. at 41,
`
`42, 55; Paper 25 (“Sur-Reply”) at 2, 17-26; Ex. 2007 at ¶¶112-113, 135-144. The
`
`Board agreed with Qualcomm: “Patent Owner argues that the intermediate storage
`
`area of Bauer and Svensson is a temporary buffer …. We agree with Patent Owner.
`
`Svensson discloses that the intermediate storage area is reserved at runtime of the
`
`program loader .... Thus, we find that the intermediate storage area of Bauer and
`
`Svensson is a temporary buffer.” Paper 30 (“FWD”) at 55-56.
`
`Intel did not challenge this factual finding on appeal, and the Federal Circuit’s
`
`opinion does not disturb it. Therefore, the Board’s finding that Bauer/Svensson’s
`
`ISA is a temporary buffer should stand on remand. Because the Board has already
`
`found that the ISA of Bauer/Svensson is temporary—i.e., not permanent—the ISA
`
`is not a “hardware buffer” as properly construed by Qualcomm.
`
`Intel tries to re-litigate the issue of whether the ISA of Bauer/Svensson is
`
`temporary, arguing that “even under a construction that excludes temporary buffers,
`
`the ISA would satisfy the ‘hardware buffer’ limitation[] since the ISA is never
`
`-16-
`
`
`
`IPR2018-01334
`
`deallocated and instead functions solely as a permanent buffer.” Intel Br. at 17.
`
`Intel’s deallocation argument is a red herring. Ex. 2015 at ¶64. As explained above,
`
`the temporary buffer described in the Background section of the ’949 patent—like
`
`the ISA of Bauer/Svensson—is temporary because it is generated (i.e., reserved or
`
`allocated) by software at run time and does not exist absent the software. Id.
`
`In Bauer/Svensson, specifically, the ISA is reserved in DSP SARAM &
`
`DARAM 108/208 by software running on the client processor 104/204. Ex. 1010
`
`at 5:3-36. Absent the software allocation, the ISA does not exist in the DSP SARAM
`
`& DARAM 108/208, and that is why it is temporary: It exists only after it is
`
`allocated by software at run time, and therefore, it is not a permanent, fixed buffer.
`
`Ex. 2015 at ¶¶65-67. The fact that Bauer/Svensson does not describe deallocation
`
`of the ISA is beside the point, and the ISA is indistinguishable from the temporary
`
`buffers described in the Background section of the ’949 patent. See, e.g., PO Resp.
`
`at 41, 42, 55; Sur-Reply at 2, 17-26; Ex. 2007 at ¶¶112-113, 135-144.
`
`B.
`
`The ISA Is Not A Dedicated Buffer Distinct From System
`Memory
`The ISA of Svensson/Bauer also fails to disclose the claimed “hardware buffer”
`
`because it is not a dedicated buffer distinct from system memory. Ex. 2015 at ¶¶68-
`
`70. Qualcomm previously showed that in Bauer/Svensson, system memory consists
`
`of the DSP SARAM & DARAM 108/208—including the portion thereof allocated
`
`as the ISA—and the DSP XRAM 110/210:
`
`-17-
`
`
`
`IPR2018-01334
`
`
`PO Resp. at 50-58; Sur-Reply at 17-26; Ex. 2007 at ¶¶135-144. The Board declined
`
`to construe “system memory” in the Final Written Decision, but Qualcomm showed
`
`that the ISA is located in system memory under both parties’ constructions of the
`
`term. See PO Resp. at 50-58; Sur-Reply at 17-26; Ex. 2007 at ¶¶135-144. Further,
`
`Intel’s declarant Dr. Lin conceded that the SARAM & DARAM 108/208 of
`
`Bauer/Svensson “could be used to load and execute programs,” such that it satisfies
`
`Intel’s construction of system memory (i.e., “memory where an executable software
`
`image can be loaded and executed”). Ex. 2008 at 53:18-24.
`
`Accordingly, the ISA of Bauer/Svensson is not a “hardware buffer” because
`
`it is allocated in system memory (i.e., the SARAM & DARAM 108/208) and
`
`th