`
`
`
`
`
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`
`____________________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`____________________
`
`
`
`APPLE, INC.,
`Petitioner
`
`v.
`
`REALTIME DATA LLC,
`Patent Owner
`
`____________________
`
`
` Case IPR2016-01737
`Patent 8,880,862
`
`____________________
`
`
`EXPERT DECLARATION OF DR. GODMAR BACK IN SUPPORT OF
`PATENT OWNER’S SUPPLEMENTAL RESPONSE TO
`ITS MOTION TO AMEND
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`TABLE OF CONTENTS
`
`
`
`I.
`
`INTRODUCTION ............................................................................................ 1
`
`II. PROFESSIONAL BACKGROUND ................................................................ 3
`
`III. PERSON OF ORDINARY SKILL IN THE ART ........................................... 3
`
`IV. COMBINATION OF SUKEGAWA, KROEKER, AND DYE ....................... 3
`
`A. A POSA Would Not Expect Any Clear Benefit From Dr. Neuhauser’s
`Proposal to Modify Sukegawa by Adding Volatile RAM ......................... 4
`
`1. The alleged cost motivation for adding RAM to Sukegawa .................. 6
`
`2. The alleged speed motivation for adding RAM to Sukegawa .............. 16
`
`B. Dr. Neuhauser’s Proposal to Modify Sukegawa by Adding Dye’s
`Compression System Would Entail Significant Implementation Costs ... 24
`
`V. COMBINATION OF SUKEGAWA, ESFAHANI, AND DYE .................... 30
`
`A. The alleged motivations for combining Sukegawa with Esfahani and Dye
`are not supported, and are incorrect ......................................................... 30
`
`B. The combination of Sukegawa with Esfahani and Dye would not meet the
`“preloading” limitation, properly construed ............................................. 36
`
`VI. COMBINATION OF SETTSU AND ZWIEGINCEW ................................. 39
`
`
`
`
`
`ii
`
`
`
`
`I, Godmar Back, declare as follows:
`
`I.
`
`INTRODUCTION
`
`
`
`1.
`
`On June 14, 2017, I submitted a declaration in support of the motion
`
`to amend and the proposed substitute claims submitted by Realtime Data LLC in
`
`this proceeding. In that declaration, I explained and concluded that the proposed
`
`substitute claims are supported by the original non-provisional application and are
`
`patentable over the prior art at issue in this proceeding, as well as the material art
`
`discussed during prosecution.
`
`2.
`
`I understand that Apple Inc. and its expert, Dr. Charles J. Neuhauser,
`
`subsequently submitted a response and accompanying declaration (Ex. 1030),
`
`respectively. I also understand that Dr. Neuhauser was cross-examined with
`
`respect to the opinions set forth in that declaration (Ex. 2024).
`
`3.
`
`On October 11, 2017, I submitted a second declaration in support of
`
`Realtime’s reply to Apple’s response. In that second declaration, I explained that
`
`the opinion expressed in my first declaration remained unchanged, and that the
`
`arguments and evidence submitted by Apple, as elucidated by Dr. Neuhauser’s
`
`cross-examination testimony, further supported my conclusion that the proposed
`
`substitute claims are patentable.
`
`4.
`
`I understand that Apple and Dr. Neuhauser subsequently submitted a
`
`supplemental response and accompanying declaration (Ex. 1043). I also understand
`
`
`
`1
`
`
`
`
`that Dr. Neuhauser was cross-examined with respect to the opinions set forth in
`
`
`
`that declaration (Ex. 2026). I have been asked to consider Apple’s arguments, Dr.
`
`Neuhauser’s declaration, and Dr. Neuhauser’s cross-examination testimony to
`
`determine whether those materials affect the analysis and conclusions stated in my
`
`declarations of June 14, 2017, and October 11, 2017. For the reasons explained in
`
`this declaration, my opinion as to the patentability of the proposed substitute
`
`claims remains unchanged, and the arguments and evidence submitted by Apple, as
`
`elucidated by Dr. Neuhauser’s cross-examination testimony, further support my
`
`conclusion that the proposed substitute claims are patentable.
`
`5.
`
`In forming my opinions, I have reviewed and considered the materials
`
`identified in the paragraphs above, those identified in my prior declarations of June
`
`14, 2017, and October 11, 2017, and the materials cited and discussed in this
`
`declaration, including the references discussed below.
`
`6. My opinions are based on my experience and knowledge of the
`
`relevant art, the documents identified above, as well as the documents discussed in
`
`this declaration.
`
`7.
`
`In this declaration, I address Apple’s references and unpatentability
`
`theories. My decision to discuss below only certain shortcomings of those
`
`references or theories should not be understood as a concession as to any aspects of
`
`Apple’s theories that I do not specifically discuss.
`
`
`
`2
`
`
`
`
`II.
`
`PROFESSIONAL BACKGROUND
`
`
`
`8. My professional background and Curriculum Vitae were provided as
`
`part of my declaration of June 14, 2017, and I do not repeat my qualifications here.
`
`III. PERSON OF ORDINARY SKILL IN THE ART
`
`9. My understanding and views as to the “person of ordinary skill in the
`
`art” were set forth in my prior declaration of June 14, 2017, and have not changed.
`
`IV. COMBINATION OF SUKEGAWA, KROEKER, AND DYE
`
`10.
`
`I understand that Dr. Neuhauser has alleged that the proposed
`
`substitute claims are invalid as obvious over a combination of Sukegawa in view of
`
`Kroeker and Dye. Ex. 1043 ¶¶ 13-43. For the reasons explained below, I disagree
`
`with Dr. Neuhauser.
`
`11. As an initial matter, I shall note that in my first declaration supporting
`
`the proposed amended claims, I observed that Kroeker had been materially at issue
`
`during the prosecution of the ’862 patent, and I explained that it did not disclose all
`
`of the limitations of the proposed amended claims. Ex. 2022 ¶¶ 66, 69.
`
`12. Dr. Neuhauser’s declaration presents a figure (reproduced here) that
`
`“show[s] how a system of Sukegawa as extended by Dye and Kroeker would be
`
`constructed.” Ex. 1043 ¶ 33:
`
`
`
`3
`
`
`
`
`
`
`
`
`
`This figure depicts Dr. Neuhauser’s proposal to modify Sukegawa in essentially
`
`two ways: (1) by “the addition of DRAM to Sukegawa’s cache system controller
`
`3” of Kroeker, and (2) a modification where “the cache system controller 3 is
`
`extended with the data compression/decompression engine of Dye.” Ex. 1043 ¶¶
`
`34, 30.
`
`13.
`
`I disagree with Dr. Neuhauser that a POSA would be motivated to
`
`make either of those modifications to Sukegawa’s system and discuss them each in
`
`turn below.
`
`A. A POSA Would Not Expect Any Clear Benefit From Dr.
`Neuhauser’s Proposal to Modify Sukegawa by Adding Volatile RAM
`
`14. Consistent with the above figure, Dr. Neuhauser testified that his
`
`proposed combination of Sukegawa, Kroeker, and Dye relies on modifying
`
`Sukegawa to incorporate the use of volatile RAM memory:
`
`
`
`4
`
`
`
`
`
`
`
`Q. In your latest declarations which we’ve been calling your third
`declarations, all of your opinions with respect to Sukegawa require
`modifying Sukegawa to incorporate the use of volatile RAM memory,
`true?
`[A.] I’m not sure about “require,” but they certainly make use of
`RAM memory, because that’s the basis of Kroeker, for example.
`. . .
`Q. And so you don’t have a theory based on Sukegawa in your third
`declarations that relies entirely on nonvolatile memory, right?
`[A.] No, I don’t think I have such a theory.
`
`Ex. 2026 at 25:5-24.
`
`15. Dr. Neuhauser’s declaration offers two reasons why he believes a
`
`POSA would have been motivated to modify the system of Sukegawa to add
`
`volatile RAM memory: cost and speed. See, e.g., Ex. 1043 ¶ 28 (“DRAM can be
`
`used to augment the memory resources of Sukegawa in a way that provides
`
`improved performance at reduced cost.”). I understand that Dr. Neuhauser
`
`presented similar testimony at his deposition:
`
`Q. And the specific reason you say that a person of skill in the art
`would have been motivated to combine Sukegawa with Kroeker is
`because, according to you, in February of 2000, the use of nonvolatile
`flash memory as taught in Sukegawa was significantly more
`expensive than the use of volatile RAM as taught in Kroeker; is that
`true?
`
`
`
`5
`
`
`
`
`
`
`
`[A.] That’s the basic motivation. There’s some other motivation that’s
`important, too, but the cost is a clear quantitative kind of thing that I
`pointed to.
`Q. What other motivation?
`A. Well, I think I -- in my declaration, I spoke about access, speed and
`write time, so it’s faster to read RAM than it is to read flash, at least in
`those days. It’s also faster to write, so there’s these access time issues.
`
`Ex. 2026 at 26:9-27:4.
`
`16. For the reasons explained below, I disagree with Dr. Neuhauser that
`
`the proposed combination of Sukegawa and Kroeker set forth in his latest
`
`declaration would present expected benefits in terms of either cost or speed that
`
`would have motivated a POSA to make the proposed modifications to Sukegawa’s
`
`system.
`
`The alleged cost motivation for adding RAM to Sukegawa
`
`1.
`17. With respect to cost, Dr. Neuhauser’s declaration states that
`
`Sukegawa’s non-volatile flash memory was “expensive” and that “one of ordinary
`
`skill would have found it obvious to make use of less costly volatile memory (e.g.,
`
`RAM)” for preloading. Ex. 1043 ¶ 13. Specifically, he alleges that “in February of
`
`2000 the cost of DRAM was significantly less than the cost [of] flash memory.”
`
`Ex. 1043 ¶ 25.
`
`
`
`6
`
`
`
`
`
`
`
`18.
`
`I disagree with Dr. Neuhauser’s assertion that a POSA in February
`
`2000 would have understood DRAM to be significantly less expensive than flash
`
`memory on a per-megabyte basis.
`
`19. The only support in Dr. Neuhauser’s declaration for his assertions
`
`regarding the relative price of RAM and flash memory in February 2000 is a
`
`citation to Dye, as he confirmed at his deposition:
`
`Q Throughout your two declarations, the only citation that you have
`to external evidence about the relative cost of flash and RAM as of
`around February 2000 is this reference to Dye, right?
`. . .
`[A.] Yes, I think that’s correct.
`. . .
`Q. All right. In your declarations, you didn’t undertake a systematic
`analysis of the literature in the late ‘90s and early 2000s to
`specifically identify the cost difference between flash and RAM on a
`per-megabyte basis, true?
`A. That’s not in my declaration, that’s correct.
`
`Ex. 2026 at 76:16-77:15 (emphasis added).
`
`20. Dye, however, does not support the proposition that RAM cost
`
`significantly less than flash as of February 2000. As an initial matter, Dye was
`
`filed in April 1999, almost a year before February 2000, which Dr. Neuhauser
`
`acknowledged at his deposition:
`
`
`
`7
`
`
`
`
`
`
`
`Q. Now, what about Dye told you what the cost of DRAM was
`relative to the cost of flash specifically as of February 2000?
`. . .
`[A.] Well, I think those statements in Dye say that the cost of DRAM
`was known, to Dye, at least, to be less than the cost of flash memory -
`- and he’s pretty clear -- on a bit-by-bit basis. Of course, this was filed
`on April 26 of 1999, so it’s a few months before 2000.
`Q. Almost a year, right, before February 2000?
`A. Right, almost a year.
`
`Ex. 2026 at 81:6-24 (emphasis added).
`
`21.
`
`Importantly, the relative cost of flash to RAM was changing rapidly
`
`between the late 1990s and February 2000. Dr. Neuhauser admitted this at his
`
`deposition:
`
`Q. You said that the only time period that matters for this
`consideration is February 2000?
`A. Well, I think that one of ordinary skill in the art has to -- you have
`to know -- you have to -- pardon me. Me, as the expert here, I have to
`put myself in the shoes of that person in February of 2000, and I have
`to say what would that person have thought at that point in time. It’s
`not February of 1999 or February of 2005. It's February of 2000.
`Q. You agree that from 1998 to 1999 to February of 2000, the
`relative cost of flash and RAM was changing?
`[A.] Yes, I agree that those relative costs were changing.
`
`Ex. 2026 at 79:5-20 (emphasis added).
`
`
`
`8
`
`
`
`
`
`
`
`22. And in fact, the relevant portions of Dye, taken as a whole, suggest
`
`that the cost of flash memory was dropping relative to RAM even before February
`
`2000, which Dr. Neuhauser acknowledged:
`
`Q. All right, and you see -- you cited to column 1, 35 through 38, of
`Dye in your paragraph 25?
`A. I did.
`Q. But you see that the sentence right before that says “flash memory
`devices are becoming more popular because of fast read and write
`response, low cost, and higher density”?
`A. Mm-hmm.
`. . .
`Q All right, but you understand that that whole passage taken as a
`whole suggests that the cost of flash is dropping, right?
`[A.] Yes, I think that’s a suggestion here.
`. . .
`Q. Well, where in your declaration do you talk about the common
`knowledge of the person of skill in the art as to the relative prices of
`flash and DRAM as of February 2000?
`A Well, I don’t think I state it in those terms, but I’m saying that this
`person certainly has this knowledge from Dye.
`Q. They have the knowledge that, according to Dye, in April of 1999,
`DRAM still costs less than flash on a per-bit basis, right?
`A. Yes.
`Q. But they also have the knowledge from Dye that the price of flash
`is dropping, right?
`A. Mm-hmm, that’s correct.
`
`
`
`9
`
`
`
`
`Ex. 2026 at 83:13-85:12 (emphasis added).
`
`
`
`23. Because the relative cost of flash to DRAM was in flux between April
`
`1999 and February 2000, a POSA would not have relied on Dye as the sole and
`
`conclusive source of information on the relative cost of flash to DRAM as of
`
`February 2000. Dr. Neuhauser acknowledged as much at his deposition, testifying
`
`that “[w]ithout any other knowledge, just based on Dye, they wouldn’t have had a
`
`certainty. . . .” Ex. 2026 at 135:6-16.
`
`24.
`
`I therefore do not believe that a POSA would have relied on the very
`
`brief and inconclusive teachings of Dye to form a motivation to incorporate RAM
`
`into the flash-based system of Sukegawa in the manner Apple and Dr. Neuhauser
`
`have proposed.
`
`25. To the contrary, there is evidence to demonstrate that flash memory
`
`could be obtained for a significantly smaller cost than DRAM on a per-megabyte
`
`basis as of February 2000. For instance, the December 1, 1999 issue of PC
`
`Magazine states that 32MB of flash-based SmartMedia memory was available at
`
`that time for $100 (retail), whereas an advertisement in the same issue shows that
`
`32MB of SDRAM cost $149.95 at that time. Ex. 2028 at 3. The same issue also
`
`states that 96MB of flash-based CompactFlash memory sold for $230 (retail),
`
`whereas an advertisement in that same issue shows that a smaller-sized 64MB of
`
`
`
`10
`
`
`
`
`SDRAM cost significantly more—$299.95. Id. at 3-4. Relevant excerpts are shown
`
`
`
`below.
`
`
`
`26. Subsequent issues of PC Magazine from the relevant timeframe show
`
`that flash remained available at lower costs than RAM on a per-megabyte basis,
`
`even as the cost of both types of memory continued to drop. The March 7, 2000
`
`issue of PC Magazine (which was necessarily prepared before March 7, 2000), for
`
`instance, shows that 64MB of flash was available for $169.11, whereas it shows
`
`that 64MB of SDRAM cost significantly more, at $249.95. Ex. 2029 at 4-5.
`
`Relevant excerpts of the advertisements are shown below.
`
`
`
`
`
`11
`
`
`
`
`
`
`
`27. Similarly, the March 21, 2000 issue of PC Magazine shows that the
`
`price of the same 64MB flash memory had fallen to $149.82, whereas the price of
`
`the same 64MB of SDRAM remained significantly higher, at $199.95. Ex. 2030 at
`
`4-5. Relevant excerpts of the advertisements are shown below.
`
`
`
`28.
`
`It is therefore my opinion that a POSA would have been aware that
`
`the cost of flash memory was dropping relative to RAM prior to February 2000,
`
`and that by February 2000 RAM could in fact be more expensive, or at least
`
`equally as expensive, as flash on a per-megabyte basis. Certainly the POSA would
`
`not believe that “in February of 2000 the cost of DRAM was significantly less than
`
`the cost of flash memory,” as Dr. Neuhauser’s declaration alleges, since the above
`
`advertisements and statements from PC Magazine demonstrate that a POSA could
`
`have easily obtained flash for a lower cost than RAM on a per-megabyte basis at
`
`that time. See Ex. 1043 ¶ 25.
`
`
`
`12
`
`
`
`
`
`
`
`29.
`
`I therefore disagree with Dr. Neuhauser that a POSA would have been
`
`motivated by the comparative cost of DRAM to flash as of February 2000 to
`
`modify Sukegawa by adding RAM to Sukegawa as taught in Kroeker.
`
`30. Moreover, beyond the mere cost of the memory itself, incorporating
`
`RAM into Sukegawa’s system would entail numerous design and implementation
`
`costs and consequences that a POSA would also need to consider before being
`
`motivated to modify Sukegawa in that manner.
`
`31. At his deposition, Dr. Neuhauser identified many of those additional
`
`costs and consequences, acknowledged that a POSA would take them into
`
`consideration before attempting the proposed combination, and admitted that his
`
`declaration did not discuss them:
`
`Q. You agree that there would be implementation costs associated
`with bringing in the RAM of Kroeker into the system of Sukegawa,
`right?
`[A.] Yes, I believe there would be some – we’re talking about a
`system where we’re going to increase the amount of, of preloadable
`memory. There would be an implementation cost to adding RAM.
`There would be an implementation cost to adding more flash.
`[Q.] Are you saying there’s an implementation cost to adding more
`flash that goes beyond simply the price of the flash?
`A. No. Well, I mean this is, this is going back to your, your questions
`of this morning, that there’s many factors to consider.
`
`
`
`13
`
`
`
`
`
`
`
`
`
`So one factor is do you have the board space, right, and what
`are the size of the packages? Is the flash package bigger than the
`DRAM package? Is the DRAM package -- remember, DRAM comes
`in different kinds of arrangements, right? And I'm not talking about
`just the chip itself, but DRAM can come in DIMMs and SIMMs in the
`year 2000.
`So board space would be an example. Power would be an
`example. I don’t know what the power differential between the two
`is, but also it depends on whether you have a flash that can work from
`five volts or whether you have a flash that has to have a separate
`power supply.
`We’re talking about 2000. It probably was available in 5 volts
`or 3.5 volts, but there’s other factors -- this is what you were getting
`at before, before you can -- you know, any design, to be realized, has
`to consider one way or the other -- if you don’t consider it,
`sometimes you get hurt -- all of these factors.
`So, you know, it’s not just a simple matter of flash is free to
`add. It’s not. There’s an implementation cost to adding it. As you said
`this morning, there’s a design cost at least.
`Q. Would a person of skill in the art, in your opinion, have considered
`the competing implementation costs aside from the cost of the
`memory itself as between using more nonvolatile flash and instead
`adding RAM in deciding which approach to take?
`[A.] It would depend on the circumstances. It would depend on the
`type of memory. They, they certainly would consider -- you know, the
`first part of an engineering process is to consider enough
`
`14
`
`
`
`
`
`
`
`information to decide whether to go ahead, and then, you know, from
`there you go ahead and solve problems.
`Sometimes you run into a dead end, right, but I think, you
`know, it would depend on the circumstances. It would depend
`whether they had board space available, whether they had certain
`types of connectors available, certain power supplies available
`already, or would they have to add a new power supply.
`All of those things they would have to consider, and those
`would be on an individual basis.
`Q. You don’t talk about those considerations in your declaration,
`right?
`A. No, I do not.
`
`Ex. 2026 at 142:23-145:24 (emphasis added).
`
`Q Well, in your declaration you don’t talk about a specific
`circumstance where a person of skill in the art has considered more
`than simply the cost differential between flash and RAM and decided
`that they can add the RAM, that it is implementation-cost-efficient to
`do so, that it aligns with their goals and purposes and technical
`capabilities and all of those considerations, right?
`[A.] I think the only ones that I discuss are the speed, the read/write
`speed and the write access speed and the -- well, it’s the same thing -
`- and the cost differential.
`
`Ex. 2026 at 146:25-147:14 (emphasis added).
`
`32. Dr. Neuhauser’s declaration does not present any analysis explaining
`
`why a POSA would have anticipated that any cost savings from purchasing RAM
`
`
`
`15
`
`
`
`
`instead of flash (of which there is no evidence) would have outweighed the
`
`
`
`significant additional design and implementation costs and consequences of
`
`modifying Sukegawa to incorporate RAM in the manner proposed. To the
`
`contrary, it is my opinion that a POSA would have viewed those design and
`
`implementation costs as significant and would have been dissuaded from
`
`modifying Sukegawa’s system to incorporate RAM even if RAM costs were
`
`comparatively cheaper than flash, and would have been particularly dissuaded
`
`from doing so in light of the reality, demonstrated above, that in February 2000
`
`RAM costs could exceed that of flash.
`
`33.
`
`I therefore disagree with Dr. Neuhauser’s assertion that a POSA in
`
`February 2000 would have been motivated by the relative cost of RAM to flash to
`
`modify the system of Sukegawa to add volatile RAM memory in the manner
`
`proposed by Dr. Neuhauser and Apple.
`
`The alleged speed motivation for adding RAM to Sukegawa
`
`2.
`34. Dr. Neuhauser also alleges that speed would have motivated a POSA
`
`to modify Sukegawa by adding RAM. Specifically, he states that “flash memory is
`
`slower to access than DRAM” and that “flash memory is significantly slower to
`
`write than RAM.” Ex. 1043 ¶ 25.
`
`35.
`
`I disagree with Dr. Neuhauser’s assertion that speed would have
`
`motivated a POSA to make the proposed modifications to Sukegawa. In particular,
`
`
`
`16
`
`
`
`
`Dr. Neuhauser’s analysis considers only the comparative read/write speeds of
`
`
`
`RAM versus flash, whereas a POSA would be concerned with the overall speed
`
`impact of the proposed modifications. In my opinion, a proper analysis would need
`
`to consider the boot speed of unmodified Sukegawa and compare it to the boot
`
`speed of Sukegawa with the proposed modification.
`
`36. The unmodified version of Sukegawa teaches loading boot data into
`
`memory in one cycle and using that data for booting the operating system in the
`
`next power on cycle. In his previous deposition, Dr. Neuhauser agreed with me on
`
`that point:
`
`Q. So Sukegawa’s teaching is that the boot data that is loaded into
`non-volatile memory in one cycle is used for booting the operating
`system in the next power on cycle; is that right?
`A. Yes. I think that’s generally correct.
`
`Ex. 2024 at 22:7-11. Thus, in the unmodified version of Sukegawa, the data to be
`
`used from the non-volatile memory during a boot cycle is loaded prior to that boot
`
`cycle. Ex. 2022 ¶ 63. During the boot cycle itself, the data only needs to be read
`
`from the non-volatile memory to be used for booting. Id.
`
`37. By comparison, after the proposed modification to Sukegawa, the data
`
`to be used from the RAM volatile memory will be both written into the RAM
`
`during the boot cycle and read from the RAM during that same boot cycle.
`
`Accordingly, the relevant comparison for evaluating speed is between reading data
`
`
`
`17
`
`
`
`
`out of non-volatile memory during a boot cycle (i.e., unmodified Sukegawa) versus
`
`
`
`doing the same while also writing data into and reading it out of RAM during a
`
`boot cycle (i.e., the proposed modification of Sukegawa).
`
`38. At his deposition, Dr. Neuhauser acknowledged that reading and
`
`writing the data in the same boot cycle would take additional time and that neither
`
`he nor a POSA could know, on this record, whether adding RAM to Sukegawa
`
`would shorten the overall boot time:
`
`[Q.] So you don’t know for a fact whether the total time required to
`boot the same amount of boot data, using Sukegawa and Dye alone,
`would be faster or slower than using Sukegawa and Dye in
`combination with Kroeker as you’ve proposed it in your third
`declarations, true?
`[A.] Well, I don’t think we know for sure or I know for sure, and I
`don’t think anybody knows for sure whether a particular situation
`would be shorter, the same, or longer, but we have good reasons to
`believe that it could be shorter and good reasons to believe that it
`could be longer. It would just depend on the particular operating
`system, the particular character of that operating system, and it
`might depend on the particular situation that you boot it in.
`
`Ex. 2026 at 50:15-51:7 (emphasis added).
`
`39. Dr. Neuhauser further admitted that the impact on boot time of the
`
`proposed modification would be subject to a variety of factors and that whether it
`
`would lead to an overall speed improvement was “unknowable”:
`
`
`
`18
`
`
`
`
`
`
`
`Q. You don’t account for all of those different possibilities in your
`declarations that you submitted most recently, true?
`[A.] Well, I don’t list those particular things, but I think in the
`declaration I pointed out that, in fact, there’s -- I don't use this word in
`the declaration, but there’s a kind of “synergy” between the
`nonvolatile memory and the volatile memory.
`Remember, in my declaration, the point of combining it was to
`reduce the cost, okay, but there’s also some synergy in the sense that
`during that period of time – and this may be where we have kind of a
`different view of this. The operating system needs data to get going,
`but it doesn’t ask for all that data in one big lump, okay?
`. . .
`So whether the total time, you know, from the beginning of
`the boot cycle to some particular point during this process of booting
`is longer or shorter in the combination is unknowable, but we do
`know it might be cheaper.
`Ex. 2026 at 48:21-50:13 (emphasis added).
`
`40. Dr. Neuhauser later confirmed, and I agree, that the relative read/write
`
`speeds of flash as compared to RAM did not change the “unknowability” of
`
`whether
`
`the proposed hybridization of Sukegawa by combining
`
`it with
`
`Dye/Kroeker would yield predictable speedup when comparing to Sukegawa’s
`
`system alone:
`
`Q. Okay. Earlier we were talking about the fact that it’s not possible
`to know definitively, based on the information that’s in your
`declarations, your third declarations, whether using some RAM in the
`
`
`
`19
`
`
`
`
`
`
`
`
`
`system of Sukegawa combined with Dye, as taught by Kroeker, would
`lead to an overall faster boot process for the operating system,
`compared to not using RAM and just doing Sukegawa plus Dye.
`Do you recall that?
`A. Yes, I recall that.
`Q. And that fact doesn’t change even if we assume that RAM is
`faster to write to and read from than nonvolatile flash, right?
`[A.] I don’t think it changes the unknowability of the -- it might
`weight it one way or the other, but it doesn’t change the overall notion
`of comparing the two.
`Q. Why does the fact that RAM hypothetically may have been faster
`to read from and write to, as of around February of 2000, not change
`the overall unknowability of the total boot time that would have been
`required?
`[A.] Let me try to answer this in two parts, because I don’t completely
`agree with your statement.
`If it’s faster, it might weight the speed, so it might make you
`think that if you use something that’s materially faster, that the overall
`time might be shorter. I mean it would give you that notion that it
`might be shorter.
`What was the other thing I was going to say?
`The unknowability of the length of time it takes to load for any
`system, really, is, is -- because the operating system doesn’t always
`load the same way every time, and it’s not a, it’s not a process in
`which the operating system is dumped into, is dumped in all at once
`into the, into the RAM of the host, okay, or, you know, transferred in
`
`20
`
`
`
`
`
`
`
`one lump. It may be transferred in little pieces, and Settsu is an
`example of little pieces being transferred.
`So the process really has a lot of aspects to it that are
`unknowable, and therefore, trying to make a comparison of time is
`going to be difficult, because you just don’t know whether it might
`be faster for one system, faster -- you know, the same system.
`The two systems we’re talking about, the flash only and the
`flash plus RAM, all other factors being made equal, like the size and
`so forth, might be faster in some circumstances for one system and
`faster for the other in a different set of circumstances.
`
`Ex. 2026 at 56:15-59:2 (emphasis added).
`
`41. Dr. Neuhauser further testified, and I agree, that based on his
`
`declaration a POSA would not know whether the proposed modifications would
`
`result in an increase in speed:
`
`[Q.] Based on what’s in your declarations, the amount of detail
`that’s in there, would a person of skill in the art as of around
`February 2000 have known whether it would take more time, less
`time, or the same amount of time to load the same number of
`megabytes or megabits or kilobytes, whatever, of boot data from
`nonvolatile flash alone, as taught by Sukegawa plus Dye, versus
`nonvolatile flash and some amount of RAM, as proposed in your
`combination of Sukegawa, Dye and Kroeker?
`[A.] Well, I think the short answer is no, they wouldn’t, but the more
`complete answer would be that they would have good reasons to think
`that it might take less time and some reasons to think that it might take
`
`
`
`21
`
`
`
`
`
`
`
`take
`it might
`think
`to
`time, probably a few reasons
`more
`approximately the same time. I think there’s good reasons to think it
`might take less, because it would depend.
`. . .
`[Q.] The answer to my question is that the person of skill in the art
`wouldn’t really know for sure either way, because, at a minimum,
`there are a number of factors that could influence the answer that
`we don’t have the details to, such as how big is the RAM, how much
`data is being stored in that RAM, and the time that’s required to do
`that, the differential of time between reading and writing to the flash
`versus the RAM and so forth, right?
`[A.] And I would add to that the behavior of the operating system on
`any particular boot cycle. Even on the same system, two different boot
`cycles might have two different times. We don’t know that without
`having more information about a specific system. There’s no
`guarantee that it’s faster or slower.
`
`Ex. 2026 at 60:9-63:1 (emphasis added).
`
`42. Finally, Dr. Neuhauser testified that not only does his declaration not
`
`provide enough information for a POSA to determine whether the proposed
`
`modification would result in a speed increase, but also that answering that question
`
`“[m]ight be a difficult engineering question” that his declaration did not consider:
`
`Q. And I think we’ve said several times -- I think you’ve said several
`times that it’s just not clear, without additional information, whether
`adding in RAM from Kroeker is going to make the overall time
`required to load boot data for the operating system from nonvolatile
`
`
`
`22
`
`
`
`
`
`
`
`flash plus RAM to be less time than simply taking the same boot
`data from nonvolatile flash.
`Is that true?
`[A.] I think that it’s true in the general sense that, without specifics,
`we don’t know, but of course, we don’t -- you can’t make a general
`statement that it would be faster or slower is what I’m saying. Might
`be the same, might be either or the same.
`The other fact, you know, that we have to consider is that the
`one of ordinary skill in the art is an adult, right? They’re an engineer,
`and this is an engineering question, right? How much RAM would
`you need to optimize this?
`Might be a difficult engineering question, might require some
`testing or some estimates, but because you don’t know whether it’s
`going to be longer or shorter, you would say to yourself, well, I better
`consider these factors and decide how much RAM would be
`important, whether it’s worth the savings in cost.
`It’s engineering. It’s not, it’s not just, you know, trying to state
`an absolute fact without enough basis for that.
`
`Ex. 2026 at 67:23-69:6 (emphasis added).
`
`43.
`
`I agree with those aspects of Dr. Neuhauser’s testimony. Specifically,
`
`the proposal to add volatile RAM to the system described in Sukegawa and modify
`
`its operation to require both writing to and reading from that volatile RAM during
`
`the same boot cycle may slow