The story about Rambus becomes very clear from Mr. Monahan's opening statement to the Virginia Court - here's the official transcript:
MR. MONAHAN: May it please the Court,
16 counsel, ladies and gentlemen of the jury. I’m David
17 Monahan. It’s my privilege to represent the plaintiff
18 in this case, Rambus.
19 This is a case about inventions. It’s also
20 a case about two brilliant inventors who have
21 revolutionized an industry, an industry driven by
22 technology.
23 It is also a case about infringement. That
24 is, using some of those inventions’ end product,
25 memory chips, while willfully refusing to pay for that
22
1 use. The inventors are the founders of Rambus, the
2 plaintiff in this case. The infringer is Infineon,
3 the defendant in this case, until quite recently part
4 of a company, a large multi-national conglomerate
5 called Siemens.
6 Let me tell you how these two inventions
7 came about. Two teachers meet for dinner. Back in
8 October 1988, Palo Alto, California, near Stanford,
9 the heart of Silicon Valley, two teachers met for
10 dinner at a restaurant called St. Michaels.
11 One of them, the one who called the meeting,
12 had grown up on a farm in Indiana. He had gone to
13 Purdue, graduated early from high school, went on to
14 Purdue, an engineering university. Brilliant there.
15 Then went on and got his Ph.D. in electrical
16 engineering at Stanford. At this time he was a
17 teacher. He was a professor at the University of
18 Illinois, and he had an idea. He had an idea that he
19 had been thinking about for sometime.
20 He basically had a vision that would
21 revolutionize how computers work. He had a vision
22 that would make them run at speeds that were then
23 unheard of - 500 million cycles per second. That’s at
24 a time when the fastest were running something like
25 20 million cycles per second.
23
1 He not only had a dream for the technology,
2 he had a dream for bringing this technology to the
3 world through licensing, but he knew he couldn’t do it
4 alone. He needed some help. And that brings us to
5 the second person at that restaurant meeting at St.
6 Michaels nearly 13 years ago.
7 Another brilliant student, another person
8 who graduated from high school early, went on to the
9 Massachusetts Institute of Technology, MIT, electrical
10 engineering, awarded his Ph.D. from Stanford in
11 electrical engineering. Also a teacher. Then, as
12 today, a full-time professor at Stanford. Is name is
13 Mark Horowitz. The first man was Mike Farmwald, the
14 inventors.
15 So Mike explained his vision to Mark, and
16 Mark was a little skeptical. He was skeptical not
17 only of the vision for this technology, he was
18 skeptical on the way that Mike was planning to bring
19 it to the world. But as he listened, and as he asked
20 questions, he began to think, Maybe, maybe we can do
21 this.
22 So starting with that dinner at St. Michaels
23 Restaurant nearly 13 years ago, these two inventors,
24 teachers, came up with these inventions and then
25 taught the world how to use them. These are
24
1 inventions that have radically changed the way
2 computers work for the last decade and may change it
3 for years to come in the future.
4 To understand what these inventions do - I
5 could use some of this technology right here - we have
6 to understand a little bit about basic components of
7 the computer. We all use them now. We saw your hands
8 go up. There is something called a CPU on every
9 computer. Sometimes that’s called a microprocessor.
10 That’s the brains of the computer that does all these
11 complicated functions.
12 There’s something also called main memory.
13 And main memory in your computer is almost always
14 dynamic random access memory. That’s a mouthful, so
15 I’m going to call it DRAM. That’s what the products
16 are in this lawsuit. That’s different from the
17 long-term memory called the hard drive.
18 This is the main memory where you load
19 things, where you type things in, where you read
20 things out on your output device, it says here.
21 That’s your display or monitor or your input device.
22 That’s your computer keys or your mouse. Isn’t
23 technology great?
24 What is a CPU, again? That’s that little
25 chip. You probably know them as microprocessors.
25
1 Intel invents them. The next one coming out and we
2 will all get one of these days is called P4 or Pentium
3 4. That’s the brains of the computer. It has to
4 store things on a short-term basis somewhere and
5 that’s where DRAM or main memory comes in.
6 There was a problem that they were trying to
7 solve. And it’s called the performance gap. What
8 that means is that the speed of these CPUs or
9 microprocessors had gone up 200 times in the previous
10 decade. They were getting very fast very quickly.
11 The problem was that the DRAM main memory that had to
12 communicate with these microprocessors or CPUs weren’t
13 getting much faster, certainly not at that rate. In
14 the same decade they had gone up perhaps 20 times.
15 So there was a gap that was growing between
16 the speed at which the CPU could run and the speed at
17 which the main memory could run. And that is where
18 the Rambus inventions were directed.
19 All of the inventions in this suit, and I’m
20 going to give you four names, it’s three patents, are
21 directed at improving the way the CPU communicates
22 with the DRAM main memory. So think of those arrows
23 like a highway. And if there’s congestion on that
24 highway, it can slow the communications down. And
25 that’s what all of these inventions were aimed at.
26
1 I am going to give you some shorthand names
2 to use for these inventions. First, they work
3 together. The inventions alone are important, but
4 it’s the inventions in combination that give us these
5 radical speeds, this improvement in speeds between the
6 way the CPU and the DRAM memories operated.
7 The first invention I want to talk about is
8 called the delay time invention. The delay time
9 invention is like putting a stoplight on each of these
10 DRAM memory chips. We don’t like stoplights. They
11 make us stop. They slow us down. They aren’t very
12 nice things, but we do know that if they are properly
13 run and properly placed, traffic overall runs better.
14 So they benefit us all. Same way with these
15 inventions. By stopping the flow of data at the right
16 times, it makes everything run faster.
17 The second one of these inventions is called
18 variable, sometimes programmable, block size. That’s
19 block size. That’s the shorthand name. And this is
20 like a metering signal on the expressway. When you
21 have an on ramp and it says stop, and then you go on
22 green. Well, sometimes they say one car go on green.
23 Sometimes they’ll say two cars go on the green.
24 Sometimes they can say three cars go on the green.
25 And that’s exactly what this variable block size does.
27
1 It says, okay, we’ll give you three bits of
2 information that can go on the green. Same kind of a
3 deal. You don’t like stopping, but this is a way of
4 improving the flow of data.
5 The third is we’ll call dual-edge clocking.
6 Sometimes called double dataing. What that does—I
7 couldn’t think of how to do it with a light, so I’m
8 going to do it with a grandfather clock. Think as
9 you’re sitting there on that expressway, instead of a
10 light, there’s a giant grandfather clock. It’s
11 ticking back and forth. And you can go—every car
12 can go on a tick.
13 All right. This particular invention,
14 dual-edge clocking says, okay, now we can have a car
15 go on the tick and we can have a car go on the tock.
16 It basically doubled the rate at which we were getting
17 traffic on the expressway. Now, that sounds simple,
18 but it’s a very complicated job in these devices to
19 make them work.
20 The fourth invention is called a delay lock
21 loop. Now, the inventors didn’t invent that. That
22 was a well-known circuit at the time. Engineers have
23 known that for a number of years, but putting a delay
24 lock loop on a chip, on a DRAM chip, was an invention,
25 and it was their invention, and what it does is it
28
1 makes everything run. All the circuits run much more
2 precisely than they otherwise would have run. And
3 people didn’t want to use them because, first, they
4 are complicated and, second, they take up a lot of
5 what they call real estate or space on the chip.
6 So it’s like the traffic metering lights on
7 the highways. If they are not set precisely to work
8 precisely right, they hurt the situation instead of
9 help it. So that’s what the delay locked loop was
10 supposed to do.
11 The thing was when these investors told
12 people they plan to put a delay locked loop or DLL on
13 every one of these chips, they were laughed at. They
14 were scoffed at. They said, “No, you can’t do that.
15 There’s a lot of reasons you can’t do that. That
16 doesn’t make any sense.” In fact, as late as 1996,
17 engineers were still telling them at some of these
18 major companies, “You can’t do that. It doesn’t make
19 sense.”
20 But the Rambus inventors knew a dozen years
21 ago what all DRAM designers know today, and that is
22 you can’t get these kind of speeds that they were
23 envisioning without the precise timing that this
24 circuit gives them. So those are the four inventions
25 in this suit.
29
1 The inventors wanted to bring their
2 inventions to the world. So they had to think of a
3 way to do that. So they came up with a business
4 model. More accurately, it came up with them. These
5 two college professors didn’t have the money to start
6 a company, to compete with the giants in this
7 particular industry. The industry that makes DRAMs.
8 It costs hundred of millions to build these plants to
9 make these products. So they knew they weren’t going
10 to bring their inventions to the world that way.
11 Instead, what they came up with is an idea
12 how they would license their inventions, and here’s
13 how it worked: They would apply for patents. Patents
14 take years to get. But in the meantime, they would
15 enter into contracts with these DRAM makers to make
16 what we call or what they called Rambus DRAMs or
17 RDRAMS, and they would license all of their technology
18 to a company such as a DRAM company, and they would
19 not only say, “Here it is,” they would work with them
20 closely. They would have their engineers work with
21 them and show them how to use their technology into
22 making these chips.
23 They would be partners. And they called
24 their licensees partners because they would work
25 together to make these products, make their inventions
30
1 work with the products. And they didn’t make any
2 money for doing that. Where they made their money was
3 if their licensees, their partners, made chips, sold
4 them to systems companies, such as Dell or Compaq or
5 Nintendo that used these memory chips, then and then
6 only would they make money. They would make royalties
7 based on those sales of DRAMS to these system
8 companies, and they would come back to Rambus. That
9 was their business model. That was their business
10 model from the start.
11 Now, the business model involved patents.
12 Here’s how patents work. They have been with us since
13 the start of our country. They are authorized by the
14 Constitution. They are a legally recognized ownership
15 right, and they are a type of property, and they are a
16 property that comes about this way: The government
17 encourages people, inventors, to make their inventions
18 public so other people can study them. They can use
19 them and build on those inventions in the future. In
20 exchange what the government does is says, okay, we’ll
21 give you something called a patent. We’ll give you a
22 patent. And what you can do with this patent is you
23 can have the right for a limited period of years to
24 exclusively use or say who gets to use your
25 inventions. And typically that’s done by licensing.
31
1 So that’s the reward for applying for a
2 patent. If you get the patent, you get some property.
3 And it’s called intellectual property. And it’s
4 property just like a house. It’s property just like a
5 farm. It’s not real property. It’s intangible
6 property called intellectual property, but it’s the
7 same in the sense that you can lease your farm for
8 rent. You can lease or license, it’s called, your
9 patents for royalties.
10 So Mike and Mark worked for months. And
11 what they did is they wrote a patent application. I’m
12 holding it in my hand. And what it disclosed was all
13 of their inventions. They put it in one application.
14 150 claims at the end of this. And they filed this
15 April 18th of 1990, and it disclosed everything.
16 Now, a patent is very much like a patent
17 application in the sense that they both have two
18 parts. One part of the patent application is called a
19 written description. The second part is called the
20 claims. Well, Mike and Mark wrote the written
21 description, and the written description was written
22 for engineers, so they can understand how to make and
23 use these inventions. And if a patent is issued, then
24 that’s what the engineers turn to. So that’s what
25 they spent their months on literally at the kitchen
32
1 table writing this patent application.
2 The Patent and Trademark Office says,
3 “You’ve got too many inventions in there. You have to
4 break them up.” So ultimately there were a number of
5 patents that resulted from this original patent. One
6 of those patents is, for example, a patent-in-suit.
7 It’s called—there are seven numbers in a patent.
8 I’ll just use the last three. It’s called a ‘263
9 patent. The ‘263 patent has the identical written
10 description as the original. There’s a few little
11 changes in the drawing, but substantially it’s the
12 same written description, the same teaching, the same
13 disclosure of inventions as in the original filings.
14 It’s 29 pages, and all but the last two pages are the
15 written description, including drawings that tell
16 engineers how to make and use that invention.
17 Now, other patents were issued as a result
18 of this initial filing that aren’t involved in this
19 suit. And one of them we’ll hear about later is
20 called the ‘703. The ‘703 has about 25, 26 pages, and
21 again, all but the last two are the same as the
22 patents-in-suit, are the same as the original filing
23 in April of 1990, and that’s the so-called written
24 description part of the patent.
25 Now, they had to, to get licenses to make
33
1 this business model work, they had to get out and they
2 had to disclose their inventions to the industry. So
3 here are two teachers on a big teaching job. They are
4 going to teach the industry about their inventions.
5 They decided they would visit every semiconductor
6 company that would listen to them and try and persuade
7 them to license their inventions. They did this
8 starting in 1989.
9 But to protect against one of these
10 companies stealing their inventions, they relied on
11 written agreements, contracts of a special nature
12 called nondisclosure agreements or, for short, NDAs
13 Before they would say anything to any one of these
14 companies, they would say, “You have got to sign this
15 nondisclosure agreement or NDA.” And they did. They
16 signed NDAs with over 60 companies in that first year
17 and a half, two years.
18 Once they signed the NDA, they would tell
19 all. They would reveal all their documents. They
20 would reveal something called their technical
21 description that had everything that was in the
22 patents and more. Some of the companies wanted to see
23 their patent applications that are confidential, and
24 they let them see them under this NDA.
25 They had presentations. The inventors would
34
1 explain this is how our inventions work, and this is
2 how you can make and build products using them. And
3 these presentations would be attended by groups of
4 engineers for these various companies, and they would
5 ask them questions. They would delight in trying to
6 stump them, but they weren’t able to stump them much.
7 And so the initial impressions, this can’t
8 happen, you can’t make this kind of a leap, started to
9 melt after they were on the road and talked to all
10 these semiconductor companies. It melted sufficiently
11 for them to sign up to the biggest in the world. They
12 originally had what they called a three continent
13 strategy. They wanted somebody in Europe. A DRAM
14 company over in Europe. They wanted a semiconductor
15 company here in the United States, and most of the
16 business was in Asia at that time, so they wanted a
17 couple of the companies over in Asia. That was their
18 initial strategy. They wanted to get these partners
19 to work with them to bring their Rambus compatible
20 products out to generate royalties that would again be
21 the only source of revenue for this company they were
22 starting called Rambus.
23 The reaction was expressed in different ways
24 when they first made their pitches. In fact, one of
25 the Japanese companies, a couple of them, were just
35
1 politely silent when they told them what they could
2 do. A Korean company wasn’t quite so subtle and
3 actually accused them of lying when they said they
4 were going to make these computers running at 20
5 megahertz at a time go 500 megahertz.
6 But finally, they won them over, and they
7 signed their first DRAM company, which was Toshiba in
8 1990. And soon it was followed by Fujitsu, and not
9 long after that by NEC. These are all giants in the
10 industry where probably less than a dozen companies
11 make almost all the supply - the world supply of
12 DRAMs.
13 So there was no mistake about using their
14 inventions improperly before patents were issued, they
15 required their licensees who had signed these
16 nondisclosure agreements to promise in the license not
17 to use their inventions for anything other than this
18 special type of DRAM called a Rambus DRAM or RDRAM.
19 And just for a good measure, they put field of use
20 restrictions in these licenses that says, “You cannot
21 use our inventions or any part of them for any
22 competing parts. Just the parts these Rambus DRAMs
23 are going to earn revenues for us.”
24 Finally, they signed up everybody. In fact,
25 even Infineon signed up in 1997. But I’m getting
36
1 ahead of myself. One of the first companies, I think
2 the first DRAM company that the inventors talked to
3 was Siemens. That was in February of 1990 before they
4 even had filed their initial patent applications.
5 What they did is they went to this company. Again,
6 it’s a large company. It has hundreds of thousands of
7 employees and does business in the tens of billions of
8 dollars a year, and had at that time something called
9 Siemens Semiconductor.
10 So it went to Siemens Semiconductor and
11 said, “We’d like to tell you about these things, but
12 first, you have to sign a nondisclosure agreement.
13 And if you sign it, then we’re going to give you
14 information that will let you decide whether you want
15 to be one of our partners. We’d like to have you as
16 our European partner.
17 Of course, Siemens signed. They signed this
18 agreement in February of 1990. And over the course of
19 the next almost two years they provided information to
20 Siemens. They thought it would take Siemens until
21 maybe April of 1990 to make a decision on whether to
22 sign a license with them or not. Siemens never did
23 sign a license at that particular time. But they were
24 interested in the technology.
25 After signing a license agreement in
37
1 February of 1990, there were a lot of exchanges.
2 There were personal meetings. There were
3 communications by phone. There were communications by
4 letter, fax, for the most part, and it was an exchange
5 of—it wasn’t an exchange, it was a one-way movement
6 of information of the Rambus interface technology.
7 Information about its inventions, all the details
8 about its inventions, how to use those inventions
9 going from Rambus to Siemens.
10 The Siemens people were generally positive
11 with respect to Rambus and encouraged them to give
12 them more and more information, but what we know now,
13 didn’t know then, is that they weren’t interested in a
14 license at the time. What they were interested in
15 from the outset was using selected portions of these
16 inventions in Siemens memory chips.
17 Here’s a letter. It’s the 28th of February,
18 1990, and it’s about a visit by Mr. Davidow, and a
19 number of people participated including a Mr. Penzel,
20 the head of Siemens Semiconductor, and Dr. Horninger.
21 Here’s what they found. Investigations with
22 regard of implementability of Rambus technology into
23 the Siemens particular size DRAM design need to be
24 continued.
25 Continuing on, further investigations need
38
1 to be performed with regard to application of Rambus
2 technology to video DRAM, not a Rambus product.
3 Here’s letter to Dr. Horninger with copy to
4 head of Siemens Semiconductor, Mr. Penzel and others,
5 March 6, 1990. “Do you think that the Rambus model
6 may be applied to our”—then it goes on to give some
7 numbers—“without changes being necessary to our
8 existing DRAM designs.”
9 So they were interested. They were
10 interested for the wrong reason. They were interested
11 in using portions of the Rambus inventions in their
12 chips, not in making Rambus products at that time.
13 Other memos widely distributed to the
14 highest executive levels in Siemens concerned the
15 positive impression they had of Rambus technology and
16 the positive interest they had.
17 After nearly six months of effort without
18 hearing anything from Siemens about a license,
19 Dr. Davidow writes a letter and says, in July of 1990,
20 he says, “Well, I’m sorry to hear that you’re not
21 interested. Or more accurately, I haven’t heard
22 anything from you for a while. Will you kindly send
23 our confidential materials, all of it that we’ve sent
24 to you, back.”
25 Immediately he got a letter from the head of
39
1 Siemens semiconductor, “No, no, we’re interested. We
2 just need more information.” And Rambus obliged.
3 It sent them their latest technical
4 description. It made presentations to them. They had
5 more personal meetings, a number of them. Everything
6 on top of that line are communications from Rambus to
7 Siemens. Everything below the line is what was going
8 on at Siemens, or at least what we know was going on
9 at Siemens at the same time.
10 We now know that the inventions in this
11 suit, particularly the delay time register and this
12 block size invention, were of particular interest to
13 Siemens. They were all over the information, their
14 engineers were. They even made drawings using parts
15 of the Rambus inventions.
16 Here’s a drawing made by a Mr. Michael.
17 They do their dates a little differently there. And
18 it’s September 23, 1990, and it shows access time and
19 block size. These are two of the four inventions that
20 I told you about. The Rambus inventions.
21 Incorporated not into a Rambus product, but
22 incorporated into a proposed Siemens design.
23 Later in 1991, Siemens brought in—it’s
24 the central research labs. The central research labs
25 for this very large company. It wasn’t just for
40
1 semiconductor division central research labs, it was
2 central for the entire company. And they said, “We
3 want you to do a further analysis of what we have
4 here, this Rambus technology.
5 So in a letter, actually a memorandum,
6 July 24, 1991, they get a report back from central
7 research labs. I’ll read it to you from here. Here
8 we go. “In recent years”—this again is Munich,
9 July 24, 1991. In recent years the access time of
10 DRAM memory chips could not keep up with the
11 requirements of modern microprocessors. The same
12 problem they were talking about back in St. Michael’s
13 Restaurant in October of 1998.
14 Then it goes on to say, Rambus is a new type
15 of approach for an efficient memory interface. The
16 semiconductor’s department has asked, and then all
17 that alphabet soup, the central research labs, for an
18 analysis of the suitability of the Rambus concept from
19 a system perspective.
20 They got that analysis. They continued.
21 The Rambus, as a memory concept, represents a
22 technological step forward. This is straight from
23 central research labs at Siemens.
24 Here’s a report they gave in July of 1991.
25 And it has a conclusion on the last page that I think
41
1 you might be able to read. I’ll read it to you. It’s
2 in the lower left-hand corner. It’s a slide
3 presentation. Rambus is technically and conceptually
4 a step further ahead but is that enough for market
5 acceptance? They were wondering about that then. And
6 in the bottom they asked the question, this is central
7 research: “Is Rambus a jewel?”
8 Now, after this years of study, 1990 and
9 1991, by the time they got to 1992 they were ready to
10 act. And they had an idea, an idea that others
11 suggested to them perhaps, and that was how could they
12 use part of the Rambus inventions in their products,
13 their SDRAM products that they were designing. And
14 what they did was they had an idea that they were
15 going to make it public domain. And here’s where they
16 got it.
17 They had a meeting back in April of 1992,
18 and in this meeting there was a report. Actually it
19 occurred on April 10 of 1992. And there was a report
20 on the meeting. The participants included some of the
21 top executives at Siemens. I’ll just tell you what
22 the report said. The report basically said—well,
23 here it is. It’s working now.
24 NEC and Samsung have announced samples for
25 the end of ‘92, and TSB, that’s Toshiba, for ‘93,
42
1 public domain version of a Rambus memory, exclamation
2 point.
3 So, this was Siemens’ view of these new
4 samples. And the evidence will show that this new was
5 because these new samples, these Sync DRAMs that came
6 out, were apparently using Rambus inventions. And
7 that’s the conclusion the evidence will show that
8 Siemens drew at that particular time.
9 So, Rambus, they say, currently no action
10 required. Preliminary results of our design study of
11 Sync DRAM show great similarity with Sync DRAM that is
12 adjacent to Rambus. Great similarity between Rambus
13 and Sync DRAM. Then they enclose a comparison between
14 Siemens and IBM positions.
15 Now, the reason that they were working with
16 IBM at that time is because they had a problem.
17 Despite its huge size Siemens had a big problem. It
18 didn’t have the design engineers to do a new SDRAM
19 design on its own. It had a number of reasons that it
20 was incapable of doing an SDRAM design on its own.
21 I’ll let Siemens executive, Mr. Bidler, explain that
22 to you in his owns words.
23 (At this time a videotaped segment of Mr.
24 Bidler’s deposition is played for the jury.)
25 MR. MONAHAN: So what they needed was help,
43
1 and they looked first to a partner. They had gone
2 into an alliance with IBM. And they were looking very
3 heavily to IBM to jointly design this new Sync DRAM
4 product with them. They worked primarily with a
5 fellow named Gordon Kelley at IBM.
6 So they got together with IBM in April of
7 1992, and here’s what they talked about. Here’s a
8 report of that telephone conference on April 29, 1992.
9 And again the participants were Gordon Kelley, Mr.
10 Meyer, you’ll hear more about Mr. Meyer, Dr. Peisl,
11 and then kept everybody else, including Mr. Penzel,
12 the head of Siemens Semiconductor, informed.
13 Rambus visited the most important in-house
14 users at IBM. Rambus is still being observed by IBM.
15 Rambus has announced it is demanding 10 million from
16 Samsung because of similarity of SDRAMs with the
17 architecture of Rambus memories. IBM is therefore
18 seriously considering purchasing a license as soon as
19 possible and as a precaution (at the introductory
20 price.) So this is what Siemens and IBM were
21 discussing about Rambus back in April of 1992.
22 At this same time, though, there was a
23 confidential report that was issued within Siemens and
24 basically there was a decision made at this point.
25 We’re going to copy Rambus. Not all of their
44
1 inventions. We don’t want all of them. We’re going
2 to copy some of them.
3 And here’s what they reported April 30,
4 1992. This is a report, April 30, 1992, and a summary
5 appears on the first page. And they are referring to
6 eliminate the memory bottleneck. Same kind of thing
7 talked about back at St. Michaels in October of 1998.
8 Next. Moving to page 2, in the JEDEC
9 Committee on MOS memories, attempts have been made for
10 roughly a year now to define a universally usable
11 successor for the previous DRAM, so-called Synchronous
12 DRAM (SDRAM).
13 Next. Samsung and NEC have announced
14 samples for the end of 1992 and Toshiba for mid-1993,
15 and they are apparently ready to enter the market
16 without waiting for eventual standardization.
17 These are the same ones referred to earlier
18 in the memo in April. They weren’t standard. They
19 weren’t interchangeable. They were going off on their
20 own with these SDRAM parts.
21 Next please. The original idea of the SDRAM
22 is based on the basic principles of a simple clock
23 input (IBM toggle pin) and the complex Rambus
24 structure. NEC (Rambus license holder) was the first
25 to suggest a leaner public domain version based on
45
1 this. Maintain a synchronous controlled two banks,
2 four-fold, internal data bus, four-word register, the
3 data output, and so on and so on, from the Rambus
4 while leaving off the proprietary Rambus control
5 protocol.
6 So they didn’t want all of it. They wanted
7 to leave some of it off. They just wanted to pick
8 some of the inventions.
9 It is possible and it is being discussed to
10 go even further and to restrict it to one bank as well
11 as two-fold data bus synchronous control. Samsung
12 seems to follow this path. That’s with their Sync
13 DRAMs.
14 In the meaning time, it’s become clear that
15 a Rambus memory can easily be converted into a SDRAM
16 (one or two banks) or conventional DRAM.
17 Next. All right. So what else was going on
18 at this time? IBM and Siemens were continuing to get
19 together and trying to work out what they were going
20 to do jointly to come up with this SDRAM product.
21 They had an alliance. An alliance Toshiba joined in
22 1992. But here it’s just Siemens and IBM. And here’s
23 what they’re saying. They’re putting together the
24 pros and cons of going with this Sync DRAM or Rambus
25 DRAM.
46
1 Under the pros, they say, for Sync DRAM,
2 it’s public domain. Anybody can use that. The cons,
3 Rambus patents. That’s the cons they were evaluating
4 in May. May 6 of 1992.
5 And for the pros under the Rambus DRAM they
6 said, well, it’s faster than a Sync DRAM, but the con
7 is Rambus property. Willie Meyer, May 6, 1992.
8 Now, what their strategy was, basically—
9 Oh, attached to that report I might add was a block
10 diagram of the proposed joint IBM-Siemens SDRAM, and
11 it had the delay time register, the access time
12 register, the same ones I showed you in the earlier
13 drawing right smack in the middle there. That’s what
14 they were planning to do. So that was their SDRAM
15 product they were playing on incorporating selected
16 parts, not all, they didn’t want all of the Rambus
17 inventions. That’s want they wanted to develop with
18 IBM.
19 Their strategy actually was to use a
20 standards group called JEDEC as a placeholder for
21 their proposal. Because they were behind the Asian
22 DRAM manufacturers, behind by a mile, and because of
23 this lack of resources that Mr. Meyer explained to
24 you, they wanted to be a player in the world market,
25 the world DRAM market. They didn’t want to lag too
47
1 far behind, although they weren’t leaders, but they
2 needed a strategy so they would be able to close the
3 gap somewhat between the Asian DRAM makers and
4 themselves at the appropriate time.
5 So what they did is they said, okay, we’re
6 going to join this effort to make Rambus public
7 domain. We, IBM, and Toshiba are going to get
8 together, and we’re going to help put what we want
9 from Rambus into the public domain. And the way they
10 did that was through a committee called JEDEC that I’m
11 going to talk to you a little bit more about later.
12 But they had some—there was some
13 disagreement. In fact, at one point they appeared to
14 be upset that they weren’t getting enough credit for
15 putting things into this JEDEC public domain standard
16 that they hoped to see. So there’s a memorandum, and
17 it’s April 1st of 1993. And the memorandum basically
18 is complaining—I’ll just read it.
19 Dr. VonZitzewitz, Dr. Beinvogl and myself
20 are somewhat astonished at the procedure that has been
21 agreed upon. I’m sure there are reasons for it. Why
22 is there not to be a joint IBM, Siemens, TBS - is for
23 Toshiba - presence? I suspect that one has given
24 priority to Toshiba as author of the first proposals
25 and Siemens-IBM are supposed to introduce any
48
1 modifications that result from in-house of JEDEC
2 discussions in subsequent JEDEC sessions.
3 All right. So they’re saying Toshiba
4 apparently ought to be the author to take the lead,
5 and then Siemens and IBM are going to clean up with
6 modifications. And then they conclude, “That would
7 perhaps not wreak quite as strongly of a cartel and
8 might be better received.”
9 So that’s what their thinking was back in
10 1993 when they were picking and choosing selected
11 parts of Rambus inventions to make this public domain
12 SDRAM. They put their plans on hold for a little
13 while because they wanted to wait and watch. They
14 wanted to wait and watch what the market did.
15 In a memo October 13 of 1993, they said they
16 are going to look and see what the market does. They
17 are going to wait and watch.
18 Reasons? Whether SDRAM remains a niche
19 product, a small segment product, or replaces DRAM
20 will only be decided during the course of 1994. Both
21 scenarios are equally likely. And they continue. A
22 compulsory situation (DRAM no longer sellable) can
23 arise in 1997 at the earliest with two years
24 headstart. I can’t read the next one. Start in
25 August ‘94 is sufficient.
49
1 So what they’re saying in ‘93 is we don’t
2 need to go ahead as quickly as we thought we would if
3 we start in ‘94. That will be plenty. We’ll be able
4 to maintain the gap behind the Asian DRAM makers at an
5 acceptable distance.
6 Okay. So that’s what they were thinking in
7 1993. And here were their plans to make portions of
8 the Rambus DRAM public domain.
9 I told you I’d come back to JEDEC in just a
10 moment, and so I will now. JEDEC is a
11 standard-setting group. There are 600
12 standard-setting groups in the United States. They
13 are not part of the government. They are private
14 groups that are basically trade associations that act
15 on behalf of their members.
16 But in December of 1991 at the suggestion of
17 Toshiba, one of Siemens’ alliance partners, Rambus
18 began attending meetings of these JEDEC Committees.
19 Now, JEDEC is basically an acronym. It’s
20 not important what it stands for, but at one point it
21 was Joint Electron Device Engineering Council or
22 something like that. It’s part of a group. It’s
23 really just a committee, but it’s part of a group
24 called EIA, a private trade association. Again, not a
25 government association. It lobbies the government.
50
1 It does standards work, gets marketing information on
2 behalf of its members. Rambus never joined the EIA.
3 But getting back to JEDEC, its members were maybe two
4 to 300 at any one time, and on the committees Rambus
5 was attending there would be maybe 50 or 60 attending
6 the these committees, the DRAM or memory committee in
7 particular.
8 The only requirement for going to these
9 committee meetings was you had to pay your dues. They
10 were held in nice places, usually Hawaii in winter,
11 and it was a good place to see what was going on with
12 your competition, and also to meet personally with
13 your customers.
14 And so Rambus went to several of these
15 different committee meetings starting in December of
16 1991. The meetings we’re open. They were
17 non-confidential. People would bring guests on
18 occasion.
19 Some of the JEDEC members, though, like some
20 of the giant DRAM makers, had a little bit more say-so
21 than others. The real say-so in JEDEC was held by the
22 council. Rambus never attended a council meeting, nor
23 did it qualify for membership. People like Toshiba,
24 Samsung, IBM, NEC, those were the people who ran the
committees and sat on the council and made the
51
1 important decisions. But I told you it was a
2 standards committee. And some of you may know what
3 standards are.
4 I brought some small standard products so I
5 can put them in my pocket. This is a flashlight
6 battery. It’s a standard product because this one is
7 made by Ray-O-Vac, but you know if you go to the
8 store, you find one made by Energizer that you can
9 switch them around and they’ll work pretty much the
10 same way. So that’s what a standard product is.
11 Since Rambus is a standard product, it took
12 great pains to make sure that whether it’s Toshiba or
13 NEC or Samsung making its Rambus parts, the consumer,
14 Dell, Compaq, Nintendo, wouldn’t know the difference.
15 They would all be interchangeable. They would act the
16 same. So that’s how—that’s a standard product.
17 Standards sometimes are set by the market.
18 Somebody gets out in front and then people follow
19 them. Sometimes they are set by committees. But the
20 great majority of committee standards don’t result in
21 products. Much less products with any great volumes.
22 In fact, here we have a committee—
23 May I step up here, Your Honor?
24 THE COURT: Anything you want to do is all
25 right.
52
1 MR. MONAHAN: Thank you.
2 Here we have a committee with Rambus first
3 attending as a guest in December of 1991, and that was
4 right after the SDRAM standard setting began. Rambus
5 noted in May of 1992 that two of their inventions,
6 burst and latency, were being talked about. And they
7 were asked to comment on their patent situation. They
8 didn’t have a