Page 1
`
` UNITED STATES PATENT AND TRADEMARK OFFICE
` BEFORE THE PATENT TRIAL AND APPEAL BOARD
` ____________________________
` HTC Corporation,
` HTC America, Inc.,
` LG Electronics, Inc.,
` Samsung Electronics Co., Ltd., and
` Samsung Electronics America, Inc.,
` Petitioners
` v.
` Parthenon Unified Memory Architecture LLC,
` Patent Owner
` _____________________________
`Case No. IPR2015-01500 (Patent 7,321,368 B2)
`Case No. IPR2015-01501 (Patent 7,777,753 B2)
`Case No. IPR2015-01502 (Patent 7,542,045 B2)
` _____________________________
`
` DEPOSITION OF HAROLD S. STONE, Ph.D.
` Washington, D.C.
` March 17, 2016
`
` Reported by: Mary Ann Payonk
` Job No. 105102
`
`TSG Reporting - Worldwide 877-702-9580
`
`1 2
`
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Page 1 of 87
`
`

`
`Page 2
`
`Page 3
`
` APPEARANCES:
`
` ON BEHALF OF PETITIONER and THE WITNESS:
` ALLAN SOOBERT, ESQUIRE
` PAUL HASTINGS
` 875 15th Street, NW
` Washington, DC 20005
` - and -
` JOSEPH RUMPLER, ESQUIRE
` 1117 South California Avenue
` Palo Alto, California 94304
`
` ON BEHALF OF PATENT OWNER:
` MICHAEL McBRIDE, ESQUIRE
` AHMAD, ZAVITSANOS, ANAIPAKOS,
` ALAVI & MENSING
` 1221 McKinney Street
` Houston, Texas 77010
`
` ON BEHALF OF HTC:
` JOSEPH MICALLEF, ESQUIRE
` SIDLEY AUSTIN
` 1501 K Street, NW
` Washington, DC 20005
`
`Page 5
`
` H. Stone
` THE REPORTER: Appearances for the
` record?
` MR. McBRIDE: Michael McBride from
` AZA on behalf of Parthenon Unified
` Memory Architecture LLC.
` MR. SOOBERT: Good morning. Allan
` Soobert from Paul Hastings LLP on behalf
` of the petitioner generally and the
` Samsung entities more specifically, and
` the witness. Along with me from Paul
` Hastings is Joe Rumpler.
` MR. MICALLEF: Joe Micallef, Sidley
` Austin, for HTC.
` MR. SKOYLES: Aidan Skoyles from
` Finnegan Henderson for LG.
` HAROLD S. STONE,
` called as a witness, having been duly
` sworn, was examined and testified as
` follows:
` EXAMINATION
` BY MR. McBRIDE:
` Q. Good morning, Dr. Stone. How are you
` doing this morning?
` A. Good morning. I'm fine.
`
`1
`
`23
`
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
` March 17, 2016
` 9:00 a.m.
`
` Deposition of HAROLD S. STONE, Ph.D.,
` held at the law offices of Paul Hastings LLP,
` 875 15th Street, N.W., Washington, D.C.,
` pursuant to Notice before Mary Ann Payonk,
` Nationally Certified Realtime Reporter and
` Notary Public of the District of Columbia,
` Commonwealth of Virginia, States of Maryland
` and New York.
`
`Page 4
`
` Appearances (Cont'd):
`
` ON BEHALF OF LG:
` AIDAN SKOYLES, ESQUIRE
` FINNEGAN, HENDERSON, FARABOW,
` GARRETT & DUNNER
` 901 New York Avenue, NW
` Washington, DC 20001
`
`TSG Reporting - Worldwide 877-702-9580
`
`2
`
`12345
`
`6
`
`78
`
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`
`23
`
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Page 2 of 87
`
`

`
`Page 6
`
` H. Stone
` MR. SOOBERT: I just want to note
` an objection for the record. We weren't
` served -- petitioner was not served with
` a deposition notice pursuant to the
` Patent Office rules. We object to that
` and reserve any relief as appropriate
` under the rules.
` MR. McBRIDE: Okay, I'll follow up
` with my office. I'm not sure what
` happened. Are we able to proceed,
` though?
` MR. SOOBERT: Sure.
` BY MR. McBRIDE:
` Q. So I guess I'll dive into the most
` important issue. Are you aware of today's
` date?
` A. Yes, I am.
` Q. And you're aware it's St. Patrick's
` Day?
` A. Yes, I am.
` Q. Are you wearing green, sir?
` A. No, but my wife's maiden name is
` Murphy.
` Q. Okay. Well, I'll leave the
`
`Page 8
`
` H. Stone
` aerospace industry and I went to research.
` That was Stanford Research Institute.
` Q. How long were you at Boeing?
` A. Oh, just a few months. I -- it was
` not a good match. I really was a researcher.
` Q. Okay. And -- and so after Boeing,
` you said you went where?
` A. Stanford Research Institute. They're
` in Menlo Park, California.
` Q. And what did you do while you were at
` Stanford Research Institute?
` A. I did research in computers of
` various kinds and published papers and worked
` on grants.
` Q. You said "computers of various
` kinds." What -- what do you mean by that?
` What -- what kind of computers?
` A. I was looking into parallel
` computers, computers with -- today, we would
` call it intelligent memory where the logic was
` built into the -- into the memory. Those are
` the two that come to mind.
` Q. Any others come to mind?
` A. Any other computers?
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Page 7
`
` H. Stone
` sanctioning to your counsel to enforce during
` the next break. That's all right. I'm not
` either, so --
` So I'd like to -- to begin just sort
` of talking about -- a little bit about your
` background. Do you mind sort of telling me --
` let's start with your education, where you went
` to school.
` A. Sure. I was an undergraduate at
` Princeton University with a degree in
` electrical engineering in 1960. I graduated
` summa cum laude.
` I received a master's and Ph.D. from
` the University of California at Berkeley, both
` in electrical engineering, the master's in 1961
` and the doctorate in 1963.
` Q. Any postdoctorate education, any
` other degrees?
` A. No -- no degrees after that.
` Q. Let's move on to your sort of work
` history. Following your Ph.D. in 1963, what
` did you do next?
` A. Well, my first position was a brief
` one at Boeing. I decided not to stay in the
`
`Page 9
`
` H. Stone
` Q. Or, yeah, any other --
` A. Any other research?
` Q. Yeah. Let's start with that. Well,
` let's start with computers. Any other
` computers that come to mind?
` A. Those were the two main things.
` Q. And then any other sort of fields of
` research that come to mind?
` A. I was working in what was called
` cellular logic, and that was a forerunner of
` integrated circuits.
` The -- the work went along a line
` that is very much like PL -- programmable logic
` arrays today. So I would say it's a forerunner
` of programmable logic arrays.
` Q. Just curious. How -- how did they
` differ? What was the -- what separates a
` programmable logic array from the cellular
` logic?
` A. Well, our -- our cellular logic was
` basically our -- our vision of what became
` programmable logic arrays. It -- it had a
` layout, a rectangular layout as they do.
` Perhaps the -- the interconnection
`
`TSG Reporting - Worldwide 877-702-9580
`
`3
`
`Page 3 of 87
`
`

`
`Page 10
`
` H. Stone
`structure that we used and the means of
`programming is -- is different from what
`evolved. We had to visualize what the future
`technology was like. And I don't know that we
`got it right or wrong; we just had something
`that's different from what evolved.
` Q. Okay. So I have parallel computing,
`intelligent memory, cellular logic. Anything
`else? I guess we can keep this while you were
`at Stanford Research Institute.
` A. That's right, right. I did work in
`design automation.
` Q. And what is design automation?
` A. Software for the development of
`computers. My design automation software did
`layouts for automatic wiring machines. Of
`course, that's -- that's now obsolete, but it's
`not -- not terribly different from the layout
`that goes on at -- at some level of integrated
`circuit manufacture today when you -- when you
`lay out the wiring between transistors on a --
`on a mask. So I was doing work that's related
`to things that -- that happen today.
` Q. Okay. Anything else you can think
`
`Page 12
`
` H. Stone
`That was on logic and memory. One was on
`software for a parallel computer. Those are
`three that come to mind.
` Q. Would it be accurate to say that your
`papers mirror the research that you were doing
`at the time?
` A. It would be accurate. There may be
`some -- some research that was not covered by
`papers published in that time.
` Q. Okay. Other than computer research
`and publishing papers, I mean, would you say
`that there was any other kind of different
`category of thing you did at Stanford Research
`Institute?
` A. I wrote grant proposals.
` Q. And what does that involve?
` A. I responded to requests for quote
`where typically a government agency was looking
`for work to be done in particular areas, and I
`was part of a team, usually the -- one or two
`or three people would get together and write up
`a response and bid for the work.
` Q. Okay. So what -- what sort of work
`were you bidding for?
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Page 11
`
` H. Stone
`of, or does that pretty much sum up in your
`mind the -- the research projects, or at least
`the main research projects at Stanford Research
`Institute?
` A. Those -- those are the ones that come
`to mind now. There may be others. I don't --
`don't mean to leave anything out, but those are
`the ones that I can recall right now.
` Q. Okay. And just -- just to fill in a
`gap, at least in my notes, what years were you
`at Stanford Research Institute?
` A. 1963 to 1968.
` Q. You mentioned that you published
`papers. Do you recall, first off, how many
`papers?
` A. It would be in my vitae. I -- I
`believe -- I can't count right now. Let's say
`on the order of half a dozen. But don't hold
`me to that number, please.
` Q. I -- I won't.
` Generally speaking, do you recall the
`subject matter of those papers?
` A. I do. One was on cellular logic.
`One was called "A Logic-in-Memory Computer."
`
`Page 13
`
` H. Stone
` A. I remember one grant request was from
`NASA.
` Q. The National?
` A. National Aeronautics and Space
`Administration.
` Q. Okay, the NASA. As opposed to Nassau
`in the Bahamas.
` A. NASA instead of Nassau.
` Q. What was -- just really quickly, what
`was that -- do you recall the subject matter of
`that work?
` A. That subject matter had to do with
`some high-level changeable software that
`they -- they were interested in software for
`spacecraft, but being able to modify it after
`launch. I think that's -- that's what comes to
`mind. I may not have that exactly accurate.
` Q. Okay. Do you recall any other sort
`of grant proposal work that you did?
` A. I know there were several. I don't
`recall offhand what they were.
` Q. So at the time of Stanford Research
`Institute between year 1963 and 1968, did any
`of your work involve video decoding?
`
`TSG Reporting - Worldwide 877-702-9580
`
`4
`
`Page 4 of 87
`
`

`
`Page 14
`
` H. Stone
` A. Indirectly, it did.
` Q. How did it involve it indirectly?
` A. I was working on something called the
`fast Fourier transform.
` Q. Would you mind taking a couple
`minutes to describe fast Fourier transform?
` A. Well, surely. Let's consider a
`picture, an image, and when you represent that
`image, you represent it as a set of numbers, as
`an array of numbers. Each number represents a
`color value. And for sake of argument, let's
`suppose our image is 8 by 8, so there are 64
`numbers.
` It's well known in mathematics that
`you don't need to use numbers for pixels to
`represent that image. You can use frequencies.
`So I'll explain it briefly, but if I can
`transform those -- those numbers, those
`integers of pixels into frequencies, I will get
`exactly the same number of numbers. I will
`have an array of 8 by 8, 64 numbers, each
`representing a frequency.
` Now let me explain what's happening.
`Focus on someplace on the wall behind me, for
`
`Page 16
`
` H. Stone
`fast Fourier transform. And you're probably
`wondering what that has to do with video
`decoding, or I'm not sure if you want --
` Q. I guess -- you're talking about like
`video image. I -- I -- what was the -- what
`was the project? Was it -- when you say you
`were working on fast Fourier transform, would
`you mind like kind of elaborating on what you
`meant by that?
` A. Yeah. The fast Fourier transform has
`long been used in signal processing and image
`processing. One of the important algorithms
`that are widely used. I was working in
`parallel processing. I wanted to build
`machines or design or know how to build
`machines that could do large amounts of
`numerical computation.
` The fast Fourier transform was an
`obvious candidate to consider for such a
`machine. So there were various algorithms to
`do this and I was looking at algorithms and --
`and machines that were particularly well-suited
`to those algorithms to do things efficiently.
` Q. Okay.
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Page 15
`
` H. Stone
`example, and you see a region. Think of that
`as your 8 by 8 region with 64 numbers
`representing those. Now erase that and look
`again and look at the edges in this room. You
`see there are edges along the walls, there are
`edges along shadows and things like that. If
`instead of looking at the pixels, the numbers
`of the colors, I told you where the edges were
`and told you where they were spaced, I could
`reconstruct that whole picture --
` Q. Okay.
` A. -- just by the edges. By knowing
`where the edges are and the relative spacing,
`I'm telling you the frequencies that are
`involved in that representation. So that's
`what the -- Fourier -- fast Fourier transform
`is a way of representing data, and I'm going to
`use this term "edges" loosely. That's just at
`a high level for you and me. It's more
`accurate to say frequencies. But instead of
`looking at the pixels themselves, I'm going to
`tell you where the edges are and then you'll be
`able to reconstruct the image. Okay?
` So that's -- I was working on the
`
`Page 17
`
` H. Stone
` A. I'm not sure if I answered your
`question about how this relates to image
`processing.
` Q. I think -- I think I -- I think it
`does. I mean, I don't think I need to ask a
`question on that. If you'd like to add
`anything, feel free to.
` A. I do.
` Q. All right, go ahead.
` A. An important aspect of image
`processing today is compression, data
`compression. And I've told you that if you
`represent a number by pixels -- I mean an image
`by pixels, you can also represent it by an
`equal number of frequencies.
` So by going into frequencies, you're
`not changing the number of numbers you have,
`but there's an important observation. If you
`look around the room and look at adjacent
`areas, they're all the same color more or less.
`There are actually very few edges in an image
`relative to the other pixels.
` When you go into the frequency domain
`where you use frequencies instead of pixels,
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`TSG Reporting - Worldwide 877-702-9580
`
`5
`
`Page 5 of 87
`
`

`
`Page 18
`
` H. Stone
`most of the frequencies are zeroes.
` Q. Okay.
` A. Okay? Because there are very few
`edges. So a big part of compression today is
`to go from pixels to frequencies, find the
`zeroes and then don't send them, just send the
`nonzeroes. And that gets you a factor of about
`10 to 1 reduction.
` Q. Okay. So going forward -- I
`apologize for kind of rushing through a little
`bit of this because I know we're only at 1968
`right now. What -- after 1968, where did you
`go next?
` A. I joined the faculty at Stanford
`University.
` Q. What did you do while you were at the
`Stanford faculty? What were your job
`responsibilities?
` A. I had several responsibilities. I
`had graduate students to supervise for Ph.Ds.
`I taught courses. I wrote textbooks. And I
`published papers.
` Q. How long were you at Stanford as a
`faculty member?
`
`Page 20
`
` H. Stone
`course in discrete mathematics.
` Q. So when you said peripherally
`related, it's -- it may not be concerning
`directly computation but fields that would be
`relevant to computing. Would that be sort of
`fair?
` A. It's relevant --
` MR. SOOBERT: Objection, form. I'm
` sorry, go ahead.
` A. It's relevant to computing, yes.
`Your question said computer architecture, and
`discrete mathematics and computer architecture,
`well, that's not a direct relationship. And
`compiler writing and computer architecture,
`that's not a direct relationship.
` Q. Okay. So -- but they're -- would you
`say that they were -- generally the classes you
`taught were relating to computing in general?
` A. Yes, they were relating to computing
`in general.
` Q. Okay. I only ask because I just want
`to make sure you -- there weren't like, I don't
`know, English classes that you taught or
`medieval studies.
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Page 19
`
` H. Stone
` A. I was there until 1974.
` Q. So from 1968 until 1974?
` A. That's correct.
` Q. Okay. And during that time, do you
`recall like how many students you would have
`taught on average? Or not I guess on average.
`Total estimate?
` A. Perhaps -- let me be in round
`numbers. I can't give you exact numbers.
` Q. Of course.
` A. Okay, perhaps 100 a year. And the
`larger numbers would be in undergraduate
`courses, smaller numbers would be in graduate
`courses so I'm adding all these together.
` Q. Okay. And was the fields
`generally -- sorry. Let me start over. Was
`your teaching field generally related to
`computer and computer architecture?
` A. It was, but I taught courses that
`were peripherally related to them as well.
` Q. What do you -- can you give me one
`example?
` A. I taught a course in compilers, how
`to write compilers and design them. I taught a
`
`Page 21
`
` H. Stone
` A. I did not teach things unrelated to
`computers.
` Q. All right. Thank you. So following
`1974, where did you go after the Stanford
`faculty position?
` A. I became a full professor at the
`University of Massachusetts in Amherst,
`Massachusetts.
` Q. And what -- and what years would that
`cover?
` A. 1974 to 1984.
` Q. And what were your
`responsibilities -- well, let me just kind of
`make it hopefully simpler. Were your
`responsibilities largely similar to what you
`did at Stanford?
` A. Yes, but in addition I started a
`video program for off-campus education at the
`University of Massachusetts.
` Q. And during the periods from 1974 to
`1984 kind of what would -- on average how many
`students did you educate a year?
` A. I can't give exact numbers. I'm
`going to say about 100 again, and if it turns
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`TSG Reporting - Worldwide 877-702-9580
`
`6
`
`Page 6 of 87
`
`

`
`Page 22
`
` H. Stone
`out that's inaccurate, it's -- I apologize. I
`just don't know.
` Q. No worries, Dr. Stone. I'm just
`trying to get a feel for -- for how --
`essentially how many students were underneath
`you in general. I mean, it's not -- specifics
`aren't necessary.
` And then following 1984, where did
`you go next?
` A. I joined the research staff at IBM at
`Yorktown Heights, the Watson Research Center.
`I was a manager of advanced architecture
`studies.
` Q. And what were your job
`responsibilities in that role?
` A. I was to do research in computer
`architecture considering problems -- long-term
`problems as opposed to short-term, which would
`be more like product development.
` Q. Do you mind elaborating a little bit
`on that?
` A. I was looking at ways to solve
`problems that would come up in the time frame
`of five or ten years, and I was not tied to the
`
`Page 24
`
` H. Stone
`systems. This was initiated by Japan, and the
`U.S. responded with its own attempt at building
`what was called fifth generation computing
`systems, and I was part of what was going on
`there.
` This was computing systems suited for
`doing cognitive kinds of calculations. And
`what I would say is that today, you can see the
`seeds of the work of that period in the IBM
`product called Watson.
` Q. Is that the Jeopardy contestant?
` A. Yes.
` Q. Just curious, for my own benefit. So
`when you say cognitive systems that led to
`Watson, so you were looking at like sort of
`artificial intelligence or just how it --
`problem solving, or what were --
` A. It was -- let's say artificial
`intelligence, problem solving. We had to take
`a close look at what the computer would be
`structured as to solve these problems. What do
`you need? Do you need lots of processors? Do
`you need lots of memory? What are the basic
`operations? There was a while I was interested
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Page 23
`
` H. Stone
`schedule of any specific project development at
`the time, that is, to -- a product development.
`So I didn't have to meet deadlines and things
`to get things ready for something to ship.
` Q. Okay. Really quickly to jump back,
`what years were you in this position?
` A. 1984 till 1994. My title changed
`while I was there to various things, but I was
`at IBM from 1984 to 1994.
` Q. And what sort of research did -- so
`you -- let me backtrack. You mentioned that
`you were doing research projects that weren't
`directly tied to actual consumer product
`releases; is that correct?
` A. That's correct, at the time that I
`joined them.
` Q. Did that change later?
` A. It did.
` Q. Okay. Let's start with generally the
`research that you were involved with. Do you
`recall sort of the subject matter?
` A. Yes. There was a broad title
`that's -- that was used at the time in --
`widely known as the fifth generation computing
`
`Page 25
`
` H. Stone
`in backtracking, so I built a model -- not a
`physical model but a mathematical model on how
`you can get a computer to undo its
`instructions. So every time you do an
`instruction, you can say undo the last
`instruction and back up. How do you build such
`a computer? And that makes sense when you have
`to do things like chess playing and game
`playing and other things.
` So that's kind of the nature of the
`research I was doing.
` Q. Were there any other sort of fields
`of research in addition to those?
` A. Yes. There's several that came up.
`I worked with a gentleman named Peter
`Franaszek, F-R-A-N-A-S-Z-E-K.
` Q. I'm sorry. F-R-A-N-A-S-Z-E-K?
` A. S-Z-E-K.
` Q. Okay.
` A. Peter is most noted for being able to
`double the capacity of disks. The disk you
`have in your computer, whatever it is, 200
`megabytes, a gigabyte, half of that is due to
`Peter because he figured out a way to encode
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`TSG Reporting - Worldwide 877-702-9580
`
`7
`
`Page 7 of 87
`
`

`
`Page 26
`
` H. Stone
`data on disks that allowed you to double the
`density.
` Q. Do you mind briefly describing what
`you remember about that?
` A. Oh, sure. The data on a disk is
`encoded by transitions, magnetic transitions,
`so you'll have a north pole, nearby you'll have
`a south pole, and then a north pole and a south
`pole. And the capacity of the disk is limited
`to how close those transitions are. You can't
`put them too close. Okay? Because you can't
`differentiate.
` On the other hand, you can't put them
`too far away because there's a -- you have to
`know how many bits are between any two
`transitions. And the way you tell the number
`of bits is when you see a transition, you start
`a clock, 1, 2, 3, 4; when you get to another
`transition, the clock stops and says there were
`4 bits here.
` So if you're too far away, you're
`going to lose track on your count.
` Q. Okay.
` A. So Peter said take arbitrary data,
`
`Page 28
`
` H. Stone
`San Jose and they agreed that this is better
`than what they had. But they determined that
`the benefit and the cost was not worth it,
`they're going to stay with their current
`technology, and they know that they can do
`better, but it's okay, they didn't want to
`spend the effort to develop and put out
`something else. So we -- we made a -- we made
`a contribution to knowledge, and it didn't get
`into the product.
` Q. Okay. It eventually -- well, let me
`ask it as a question. Did it eventually get
`into products?
` A. I don't know. I know it's in the
`Technical Disclosure Bulletin so that it's free
`to anybody to use.
` Q. So other than -- so we talked about a
`couple things. Were there any other sort of
`fields of research that you were doing at IBM?
` A. Yeah. There's several.
` Q. Do you mind sort of describing
`briefly some of those or listing them out?
` A. Yes. One was the statistics of
`references to cache memory and understanding
`
`Page 27
`
` H. Stone
`whatever it is, and recode it. I'm going to
`recode it in such a way that I get my -- I can
`decode it and get my data back, but in my new
`code, there are no transitions too far away.
`Okay?
` Q. Okay.
` A. And the -- when you have two
`transitions close to each other and you can --
`you make that as close as you can on the disk.
`That's determined by the physics of the disk.
`And by using this code he was able to double
`the number of bits that you can store on the
`disk.
` Q. Okay. Next time you talk to
`Mr. Franaszek, thank him on behalf of all of
`us.
` A. Okay. Well -- so Peter and I worked
`on later versions of coding and we developed a
`system for coding along the same lines. It was
`more sophisticated, and I believe we had a
`disclosure on it. We do not have a patent on
`it. It's a Technical Disclosure Bulletin that
`contains it.
` We talked to the disk people at
`
`Page 29
`
` H. Stone
`how programs and cache interact. And we used
`that both for cache memory in processors and in
`disk memory that had disk caches. There are
`papers published in that area.
` Q. And what would be sort of the -- I
`mean, if you could kind of synthesize the --
`the general problem that you were trying to
`research? Is there -- is there a problem
`statement?
` A. Well, if -- the general problem would
`be to -- understanding the statistics of what's
`going on to help you design to make you say
`it's bigger cache or smaller cache or other
`properties. I'll give you an example of a
`paper that was highly useful. It's called
`"Footprints in the Cache."
` Q. Like footprints in the sand?
` A. "Footprints in the Cache." And I
`published this with a student whose name is
`Thiebaut, T-H-I-E-B-A-U-T, Dr. Thiebaut. This
`came about because of some consulting I was
`doing with Sperry Univac, and they had a
`problem where they were designing navigation
`systems for submarines.
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`TSG Reporting - Worldwide 877-702-9580
`
`8
`
`Page 8 of 87
`
`

`
`Page 30
`
` H. Stone
` They built the computer, they built
`the cache, they built the software. Everything
`met the specs. Navigation software could
`navigate in real time, the cache could hold it,
`the computer was fast enough. But it failed
`its navigation requirement at sea trials. And
`they did not understand why.
` And so I came in and modeled the
`situation and I discovered that they -- their
`navigator software worked fine as long as it
`was in cache. It came in, ran the navigation,
`then stopped, and then other programs came in
`and they reloaded the cache with themselves.
`So the next time the navigator ran, it wasn't
`in cache.
` Q. Where was it?
` A. Well, it was in main memory and you
`had to reload the cache. And the time it took
`to reload the cache caused it to miss its
`navigation deadline.
` So this was a real time problem. I
`determined what the problem was, I determined
`what the solution was. It was simple: Add
`more cache. But I had to verify that the
`
`Page 32
`
` H. Stone
`the cache was used by something else?
` A. Yeah. As soon as something else
`comes over, it says I want to load my stuff,
`and the computer automatically brings it into
`cache and it evicts something else to make room
`for the new stuff.
` Q. Would you characterize this as a
`problem associated with, like, shared memory or
`shared resources?
` A. It would be a problem that would be
`shared resources, shared memory, real time.
`All those factors.
` Q. And do you recall the -- the time
`frame within your 1984 to 1994 period that you
`were working on this?
` A. Well, if you have my vitae -- I don't
`have it here -- I'll give you the date of the
`paper. And I continued to work on it. If I
`had my vitae in front of me, I would be able
`to -- to handle that.
` Q. Well, do you recall what that -- what
`that paper was called? Was it -- was this the
`Footprints in the --
` A. "Footprints in the Cache" was the
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Page 31
`
` H. Stone
`amount of cache that they added would be
`sufficient so that the navigator would still be
`there when it took over the machine again.
` Q. Okay.
` A. When they went to sea, it passed the
`sea trials. And as far as I know it's still in
`use, because they don't change that very much.
` Q. So -- okay. So just let me sort of
`see if I got the thing. There was a -- there
`was software trying to run in the cache and
`then that cache was being used by other
`devices?
` A. Yeah, this -- the cache is
`automatically loaded in the -- whenever
`somebody wants to -- data or instructions, it
`calls for it from memory. The computer looks
`first at the cache for it. If it's not there,
`it takes it from memory and puts it in the
`cache. And you pay a penalty to get it there.
`And that penalty had not been computed -- had
`not been part of the analysis that Sperry did
`when they -- when they proposed building the
`navigator that way.
` Q. And it wasn't left in cache because
`
`Page 33
`
` H. Stone
`first of them --
` Q. Okay.
` A. -- but I had other papers describing
`disk memory. I'd be happy to give you the
`dates and the titles and things.
` Q. Okay. I'll look for "Footprints in
`the Cache." I don't have the CV in front of me
`right now, but we can maybe turn back to it at
`the end. I don't want to eat up too much time
`but I want to get through what we need to.
` A. Surely.
` Q. While you were at IBM did you do
`anything involving video decoding?
` A. I continued to work on coding. This
`was the -- with Peter Franaszek. So that is
`related to the extent that his codes were used
`in memory. I also worked on codes related to
`what he called the 810 code which is used for
`fiberoptics. And that -- that's how two
`devices communicate over a fiberoptic channel.
`It's also used for CDs where you wo