#99 – An Interview with Steve Leibson - Impavid Ideopraxist Insider

Chris Gammell: This is the Amp Hour Podcast, recorded June 10th, 2012. Episode 99, with guest Steve Liebson, Empathed Ideopraxis Insider.

Dave Jones: Welcome to the Amp Hour. I'm Dave Jones from the EEV blog.

Chris Gammell: And I'm Chris Gammell of Chris Gammell's Analog Life.

Steve Liebson: And I'm Steve Liebson, live from Silicon Valley. Hey, Steve. Good to have you. Hey, Dave. Thanks for joining our show.

Chris Gammell: My pleasure. Thanks for asking. And Steve, you're of the 360, or is it EDA360 blog? Is that the main place to find you these days?

Steve Liebson: Well, there's two places. I write a blog every day called the EDA360 Insider, which we'll probably parse up for your listeners a little bit. And I also write the Denali Memory Report, which covers DRAM and flash and solid-state disk drives and everything else related to memory, which is a lot of fun. Yeah.

Dave Jones: So what exactly are you doing these days?

Steve Liebson: Well, I'm a marketing director for an EDA company called Cadence. Uh-oh. Marketing, folks. Marketing. Yeah, that's right.

Dave Jones: Make sure you get your bingo sheets out.

Steve Liebson: It's funny you should say that. Lately, I've run into an awful lot of antipathy towards marketing people. Unthinking ill feelings, not realizing that a lot of these people are actually engineers trying to do good.

Dave Jones: And you are a former engineer, which I'm sure we'll talk about. Former engineer? Is that... Or you still think of yourself as an engineer?

Steve Liebson: Always an engineer.

Dave Jones: Excellent. That's what we want to hear. Well, the good ones anyway. The bad ones just go on and never look back, right? They sell shoes, yes. Yeah. Okay. Got it. Well, maybe we should start there because you started out at Hewlett-Packard.

Steve Liebson: That's right. When was that? And there's a long... That was before they invented electrons. Right. Here's what happened, and it's a good story. So I graduated from Case Western Reserve in Cleveland in 1975. Come on, Chris.

Dave Jones: Way to say it. Woo-hoo!

Steve Liebson: Yeah, that's right, Chris. Sorry. Yeah, that's Bill Rust Belt. And for my first couple of years at Case, I had really, really terrible experience with computers because Case was running computers with batch systems and punch cards back then. And that was... That's a terrible experience. That is as non-interactive as you can get. Yeah. You go up to the second floor of the computer building. You sit at a key punch. You type out your program one line per card until you have a deck. Then you go downstairs to these giant reader printers that are about the size of a small SUV. You throw your cards in the hopper. You push the go button. And if Case had bought enough time on the machine during that hour, it would actually read the cards. And then after it read the cards, you would wait about five minutes. And then you would get a list of three dozen error messages where your program had bombed. And then it was time to go back upstairs to their key punches and repeat the cycle. So you can see why I had very little appreciation for computers at that point. Yeah. But when I was a junior, Case got a hold of an incredible machine. So this is 1974. It got a hold of a Hewlett Packard HP 9830 desktop calculator. Ah, nice machine. Yeah. Yeah. Well, it's full keyboard, single line LED display with alpha numerics. And the critical thing was, A, when you turned it on, you could type basic into it immediately. And B, when you shut the door, the machine was all yours. And that is the key. It was a real personal computer. And I grew so excited about that machine that I said, A, I'm going to go work for Hewlett Packard. And B, whatever division makes this machine, I'm going to go work there. Awesome. And the next year, who should come to interview at Case but a gentleman from Hewlett Packard. And unbeknownst to me, he had designed the processor in the HP 9830. So I went through the whole story again. And I said, I don't care where it is, wherever they make that machine, whoever designed it, I want to go there. Wow. Brilliant. And it turned out to be the middle of nowhere, Loveland, Colorado. Right. Now the home of Agilent, correct? Well, yeah. The Agilent's, well, the Colorado location was the first location outside of California for Hewlett Packard. And it was originally the Loveland Instrument Division. And they made mostly audio signal generators. And then they got into volt meters. And when Hewlett Packard's first calculators were looking for a home, and we're talking desk calculators here, not the pocket calculators. They ended up in Loveland, Colorado. And so that's why they started making those machines there. And Hewlett Packard left Loveland, Agilent held on, and they may still be there. But mostly I think they've moved up the road a few miles to Fort Collins.

Dave Jones: Yes, I believe they have moved. Why did they move there in the – why did HP open a – there in the first place? Yeah. Is it low-cost labor?

Steve Liebson: Well, no, it wasn't. Yeah. Lowland, Colorado was a sleepy little retirement community whose main production was sugar beets and sugar before Hewlett Packard showed up. And Packard had grown up in Colorado, and he wanted to cite his first division outside of California in Colorado. But he wanted it to be Boulder because Boulder's got a university, and Boulder had what was then called the National Bureau of Standards, now called the National Institute of Standards and Technology, or NIST.

Dave Jones: Yep.

Steve Liebson: And so they were kind of looking to cite a division there. But the mayor of Loveland and a large real estate developer in Loveland heard about Packard coming to look over Boulder and somehow got down to Boulder and snagged him and dragged him up to Loveland and convinced him to buy a large parcel of land on the south part of Loveland, which is where he cited that division.

Dave Jones: Interesting. Interesting. So he just wanted to go back to his hometown, really. He wanted to help him out. Well, it wasn't his hometown. Well, but his home state.

Steve Liebson: He actually grew up in southern – yeah, his home state. That's right.

Dave Jones: Yep. And, yeah, I can dig that. You know, he's got, you know, feelings for his home state and wants to give back, I guess, in terms of, you know, generating jobs and things like that. I'm sure it's a big deal for him.

Steve Liebson: He's a sharp guy. I suspect that he figured out that land was cheaper there and he could do well.

Dave Jones: Yep. So I'm sure it was a bit of both.

Steve Liebson: Yeah.

Speaker ?: Yeah.

Steve Liebson: So at any rate, I became a design engineer there. Right. I was responsible for designing entire products. And that was what was really unusual back then is that products were small enough and simple enough that one electrical engineer could do the whole product. That wasn't universally true, but it was true for the ones that I worked on.

Dave Jones: Right. And what products are we talking about? Any famous ones we know of?

Steve Liebson: Not a single one.

Dave Jones: Right.

Speaker ?: Okay.

Steve Liebson: But I did work in association with two very famous products. And the first one was the HP 9825 desktop computer. Mm-hmm. And that was really the computer that ignited the automatic test equipment industry.

Dave Jones: Yes.

Steve Liebson: Because it was very easy to interface to all kinds of equipment. It had an excellent language called HPL or high performance language that was sort of like basic in Polish with all the vowels sucked out. So that, you know, print was PRT instead of print. Yeah. But it was an algebraic language and it was fast. It was interpreted, but it was fast. And the Loveland Instrument Division and many other divisions at Hewlett Packard adopted it. Mm-hmm. So I designed several interface products for that machine. And I also designed a language extension, which was Pure Software Project, called the Systems Programming ROM, that extended the HPL language with some useful things for writing programs to control automatic test equipment. Right. And the other machine I was associated with was the HP 9845. Mm-hmm. And that was a full-blown desktop computer with a CRT display, two tape drives because we hadn't gotten into disk drives yet. Yeah. Yeah. Digital tape drives, DC100 cartridge tape drives, big keyboard. The machine was – you could barely lift it with one person. It was a big sucker. Tank. Yeah. It was a tank. And I worked on the IO backplane for that machine too.

Dave Jones: Nice. Nice. And that started your love affair with the old HP – the HP computers and the HP calculators.

Steve Liebson: Yes. That's – I – really, I should say, it started my love affair with personal computers. Right. The machines that were under your complete control. No timeshare. Mm-hmm. No signing up for it. It was just your machine. And that concept really excited me. And obviously, it's a concept that took off. Exactly.

Dave Jones: And HP missed the boat on that because they were probably in a very good position to come out with a personal computer, but they didn't. Is there any story behind that? Do you know any reasons why they didn't?

Steve Liebson: Well, first of all, we have to note that as of 2012, HP is the largest personal computer manufacturer on the planet. True.

Dave Jones: But at the time, yep.

Steve Liebson: At the time, yes. HP – well, here's what happened. Now, HP was adept at developing its own standards. Right. Its own designs, its own user interfaces. And when IBM came out with a personal computer in 1982, it really wasn't as powerful as the things that Hewlett-Packard had been developing. So the technology guys at HP said, you know, we're not going there. And, you know, then it became apparent that the PC architecture was going to be a runaway success. Mm-hmm. And so HP eventually did get into the fray and did start making machines that conformed to the IBM PC specification. That's right. But it did take a while to make a course correction.

Dave Jones: Yep. A course correction to jump into the pool. So what was your first piece – what was your first personal computer apart from a HP, like one that we've heard of?

Steve Liebson: Some of your listeners may have heard of what was known as the Sol 20. Yep. Okay, so you know what a Sol is. So that was my first machine.

Dave Jones: I know my computer history, yeah. So I've heard of it. I'm a young one.

Steve Liebson: The Sol 20 was a machine that was an S100 system. So it was based on the Z80 or, excuse me, the Z80.

Dave Jones: The Z80, thank you very much. Yes, you're welcome.

Steve Liebson: Got to get those Aussies. And its significant feature was that it came with a terrific keyboard made by Cherry, the highest quality keyboard in the business. And the side panels were made of oiled walnut. Ooh. So, yes.

Dave Jones: They just don't do that anymore.

Steve Liebson: They don't do that anymore. And it was a very nice S100 computer. And I got it for a consulting job where I used the entire machine to emulate a microcontroller. So this was an early version of in-circuit emulation where a basic program was taking the part of a machine code program running on a microcontroller in a telephone design for handicapped people.

Dave Jones: Nice. What microcontroller was that back then?

Steve Liebson: Which microcontroller did they end up using? The project never really got off the ground as far as I know. So it never converted over to a microcontroller. Oh, okay. It was a proof of concept sort of thing. And the company that I consulted for did not end up ultimately getting funded for that, I don't think.

Dave Jones: Because that would have been the sole. That would have been around about 78 maybe. Yep.

Steve Liebson: Right around 78. There you go. And I had a North Star floppy disk controller plugged into the S100 bus. Mm-hmm. And again, for your listeners, that used 5.25-inch floppies. But significantly, it used 10-hole hard sector floppies. Because North Star had actually designed their own floppy disk controller out of discrete TTL chips rather than use like the Western Digital controller. Yep. And it was the most bulletproof controller on the market at the time.

Dave Jones: Because they physically had holes in the disk, right? That's right. To actually align them up. And then they'd have an optocoupler in there.

Steve Liebson: Yeah. Yep. Yep. So you know exactly where it is. You could get 10-hole hard sector disks. You could get 16-hole hard sector disks. Or you could get soft sector disks. And they were not interchangeable.

Speaker ?: No.

Dave Jones: And what sort of capacity were we talking about then?

Steve Liebson: Oh, originally they were like 90 kilobytes.

Dave Jones: 90. Whopping 90 kilobytes.

Steve Liebson: Yeah. 90 kilobytes. And then we got double-sided and it was 180 kilobytes. And it was like, woo-hoo. Yes.

Dave Jones: Yeah.

Chris Gammell: I'm sure Chris is just sitting there shaking his head. I'm just sitting here. No, I'm not shaking my head. I'm just, you know.

Steve Liebson: Yeah, thanks, Mr. 2001. Yeah, right? Yeah. Yeah.

Chris Gammell: If listeners don't know, I also went to Case and I told that to Steve. And he, I shouldn't have told him. I feel bad.

Dave Jones: That was exactly 25 years between them, folks. In terms of graduating. Yeah. Oh, boy. Yep. Young whippersnappers. Ah. That's right. I don't know. Yeah. So, what did you think when-

Steve Liebson: David, you can't go too far. You told me how old you are, too.

Dave Jones: Yeah, I know.

Steve Liebson: I mean, let's not go there.

Dave Jones: So, what did you think when Agilent spun off from Hewlett Packard? Did you think that, you know, that was the best thing? And should they have kept the HP name? Or what? What did you think of that whole shakeup?

Steve Liebson: Well, I'm of multiple minds here. First of all, yes, I do think that they needed to separate. Because the instrumentation part of Hewlett Packard was having a tough time getting any attention. They were the flea on the tail of the dog, frankly. Yeah. Printers and printer ink are the things that drive HP profitability. Oh, yeah. And the cost structures of the instrumentation guys are just so different from what the printer and even the personal computer guys are doing that it's like running two different companies. And if you've got one management team trying to run two different companies, then there's going to be problems. And so, it was clear that they needed to separate. Now, as far as the name going, sure, most people who worked for HP for a long time felt that the heart of the company was the instrumentation stuff. That's where HP started, and that's where the name ought to go. But the reality is, once again, you can't give the name to the flea on the tail of the dog. You've got to give the name to the printer guys because it's a lot more revenue. It just is. And so, that's how it ended up.

Dave Jones: Unfortunately, but yeah, that's the reality, I guess. How do you split a company like that, I wonder?

Steve Liebson: Well, HP was very divisionalized. Well, HP was highly divisionalized. Right. So, you just split it along divisional lines. It's, you know, the instrumentation lines were pretty much always separate. Yeah. And so, what goes where? There's a little bit of a problem. For example, where does HP Labs go? Right. Yes. That was one of the things that crossed over the line. So, now there's, at least there used to be Agilent Labs, and I suspect there still is.

Dave Jones: Mm-hmm. I still call things HP. How about you? I used to do that, but I got corrected enough so that I don't say that anymore. Okay. I still.

Chris Gammell: And I wasn't allowed to say either, really, when I was working at Keithley, so.

Dave Jones: Ah, Keithley! Hey, you got bought out by Denner, whoo-hoo!

Steve Liebson: Well, and LaCroix just got bought out by Teledyne. Yeah, exactly. Yep. Talked about that last week.

Dave Jones: Yep. Oh, did you? Oh, great. Yes, yeah. That was our breaking news last week. Breaking news a week late.

Chris Gammell: Yeah, exactly. We're really up to stuff. Ah, well. So, what about post-HP, though? I mean, like, you've kept going. Is that when you kind of jumped over? I saw you jumped in the management role. I was kind of stalking LinkedIn for you.

Steve Liebson: No, stalking me.

Chris Gammell: Yeah. Stalking. Yeah, I didn't have your password. I didn't steal passwords like whoever those hackers were, but I did look on LinkedIn. Okay.

Steve Liebson: Yeah, I became a manager at HP. I became a very young project manager, but then I left HP shortly thereafter and joined a couple of early CAD companies. One was an architectural CAD company. That didn't last very long. They couldn't really get their act together. But the second one was Cadnetics, and Cadnetics was an early PC board CAD company. And that was in the early 80s, in the day when you had to actually design your own workstation in addition to writing the code. Oh, wow. Because there were no standard machines capable of actually running the kind of code that we wanted to run. Mm-hmm. So I ended up- But that was only for a short time, though, right?

Dave Jones: Yes, it was. Because there was just- Okay, so there was a very narrow window there of- That's right. And I guess anyone who jumped on that ship jumped on the wrong ship, right?

Steve Liebson: No, no, no, no, no. None? None? No, the first half of the 80s was like that. And so the first wave was the Daisy-Mentor Valid, or DMV wave. And Daisy, well, they're gone.

Chris Gammell: Yeah, but some of us still deal with their designs. I will say that. Daisy and Daisics, those are all still around.

Steve Liebson: Yeah, well, Daisics is a contraction of Daisy and Cadnetics, where I went. Interesting. So there was this horrible financial transaction that ended up smooshing Daisy and Cadnetics together, and it was a fiasco, which would be the polite word of the word I'm thinking of. But I was just at the design automation conference last week talking to the CEO of Mentor, who said, you know, I think we finally ended up with all of the Cadnetics tools- Oh, wow. Right. Yeah. And so their PC board stuff is at least related to, if not somewhat based on the old Cadnetics tools 20-some-odd, 30-some-odd years later.

Chris Gammell: Wow. So is that like a game of hot potato? Like, you take it. No, you take it. No, you take it.

Steve Liebson: No, it's the game of acquisition in our industry, which is sort of like the Game of Thrones on the HBO. Nice.

Chris Gammell: Well, hopefully a whole lot less incest, but you never know.

Steve Liebson: No, not even slightly less incest. No? Okay. All right. You know, if anybody that you've worked with in this industry, you're likely to work with again through some sort of acquisition.

Dave Jones: Yeah.

Chris Gammell: Okay.

Dave Jones: It's just the way it is, unfortunately. Yeah. I'm just waiting for out him to be acquired. Yeah. It's inevitable. Yeah. Oh boy. Maybe we shouldn't go there.

Steve Liebson: Maybe. Yeah. So, at any rate, when I was at Cadnetics, I was not a manager. Right. And then I switched over to journalism. And why?

Dave Jones: Why was that what made you? Was it a conscious decision or did it just something that organically happened?

Steve Liebson: No, it was a conscious decision. Right. What had happened was that when I was at HP, I started writing because my boss, Cal Finn, his girlfriend was the woman who published Keyboard Magazine for our division. That's a magazine that we sent to everybody who bought one of our machines. And I wrote a series called Liebson on I.O. that explained the different interfacing technologies. This was back in the day when you actually had to describe what is parallel I.O. What is serial I.O.? What is HPIB? And so I did that. I wrote these 12 articles. Well, Byte Magazine picked it up, ran it as a six-part series. And then Instruments and Control Systems, which was a trade magazine, picked it up and turned it into a course. So you could actually graduate from my course in I.O. Nice. Really? Yeah. And from that, I found out that I like to write.

Chris Gammell: So do some of our listeners possibly have a Steve Liebson certificate? A certificate?

Dave Jones: I wonder. We'll have to ask on the forum as well.

Steve Liebson: It's certainly possible, but I severely doubt it. So at Cadnetics, in three years, I designed five workstations. And I'll tell you, that's kind of grueling. I designed two workstations, one workstation based on the 68000, two on the 68010, two on the 68020. Wow. And by the end of that, I was pretty much done. I was designed out. And at that point, EDN Magazine ran a full-page ad. And I'd been reading EDN Magazine since high school. That's how much of a nerd I was. Right. Yay! Yeah, there we go. And it said, we're looking for a technical editor to be based in either Colorado, Texas, or Chicago to cover the regional duties of a technical editor for EDN. Mm-hmm. So I got really excited about that. I wrote off to them, and they sent me a test. And I worked really, really hard on it because I really wanted to compete for that position. And I sent it in. And they had me come out and interview. And then they finally picked me from all the candidates. And I later found out that I was the only candidate. The only candidate. Yeah. So I got the job. Whatever works. Yeah. Yeah, whatever works. And I got Texas and Colorado and Chicago because there was only one person in the middle of the country anyway. Mm-hmm. So that's how I started with EDN. I had already done a fair amount of technical writing. I enjoyed it. And things really took off at EDN. And eventually, I became chief editor of EDN. And how long did that last for? Well, I worked for EDN from 1985 to 1996. So that's 11 years. Wow. And then that company started out as Connors Publishing when I was working for them originally. It eventually became Read-El-Severe, which is a British-Dutch conglomerate. And I went back to Read-El-Severe for three years to run the microprocessor report in the early 2000s. Right.

Dave Jones: So what do you think about the magazines these days? Do you think they have any clue what they're doing with the web and all the digital content stuff and things like that? Or are they still stuck in the Stone Ages?

Steve Liebson: Well, the internet and Google have pretty much destroyed the trade magazine business. Yeah. Right.

Chris Gammell: It was probably an actual career move back then, right, to go from engineering into writing where there was actually probably some decent money, right? It was.

Steve Liebson: But it was extremely unusual because there weren't that many trade magazine writing positions for engineers. There are a lot of trade magazines. Interesting. But most of them don't employ engineers. So it was – the cadre of people in that kind of position was never more than 100 and certainly less for the most part. So it was a very rarefied atmosphere. But what's happened with the internet is that companies can now much more effectively turn into their own publishing company and get much more information made available at their whim to potential customers and prospects. And at the same time, Google pretty much sucked up all the advertising in the whole world. Yeah. And none of us saw it coming. I mean, when Google first came out, it was a search engine. And none of us could actually even figure out how they were going to monetize themselves. And what we didn't see, even those of us who were – whose lifeblood was advertising where we make all of the money in a trade magazine, none of us saw that there was going to be this black hole called Google that was going to suck up all the ad money in the universe.

Dave Jones: Yep.

Steve Liebson: So –

Dave Jones: Right.

Steve Liebson: You know, the trade magazines that exist today, and there aren't very many, they're in survival mode. They're trying to figure out what can they do to provide value to their advertisers. And it's a very, very tough thing. Yeah. I wouldn't – I don't envy them their position.

Dave Jones: Although based on some of the money that I know a lot of the companies are throwing at these websites like EDN and things like that, I would – I still like to think that there's a lot of money to be had there. These companies still seem to think that they have to be on these, you know, websites and things like that. So – Well, for some specific types of products, it does work.

Steve Liebson: Yeah. I mean, you know, Dave, you're a perfect example of where some of that money is ending up going, right? You've got test and measurement advertisers who are now advertising on your blog. That's right. That's money that's not going where it used to go, which is like ninth-page ads in the back of EDN magazine. Mm-hmm.

Chris Gammell: I wish I could get some of the – It's like 20 ads per page kind of thing where it's all blocked out and everything like that.

Steve Liebson: Well, ninth page. One – you know, nine ads per page. It's a three-by-three matrix. Oh, okay. They were very popular for companies like the Rigols of the world who say, you know, buy a 20-megahertz digital oscilloscope or a 50-megahertz or 100-megahertz digital oscilloscope. Only $7.95. Yep. That's perfect for a direct response ninth-page ad in the back of the magazine. Mm-hmm. But if you've got Dave Jones stripping one apart and showing you how it works and you've got an eyeball glued to you for 90 minutes or 20 minutes or 50 minutes or whatever it is that he decides to do, that's –

Chris Gammell: Well, most people don't watch Dave's whole video, but yeah.

Dave Jones: Yeah, they do. Unfortunately, they don't pay a linear increase in money in terms of the times the eyeball's looking. So I wish I was getting some of the money that EDN are getting.

Steve Liebson: You may think that. Dave, you need to charge one price for the unboxing video and another price for the taking it apart video.

Dave Jones: Well, at the moment, I don't charge anything. Maybe I might get a free scope out of it if I'm lucky.

Steve Liebson: Well, I think I'm doing it wrong, right? Yeah, that's all marketing. We can talk about that later.

Speaker ?: Okay.

Dave Jones: I think you've been telling me that for a couple of years that I'm doing it all wrong. I know. You still won't listen.

Steve Liebson: Yeah, I know.

Chris Gammell: Listen to Steve, Dave. He's been in the industry for a while.

Steve Liebson: I'm just trying to make you money.

Dave Jones: Oh, boy.

Chris Gammell: I know.

Dave Jones: So you went to the – sorry. Go, Chris.

Chris Gammell: Oh, well, it was the jump back then. So, I mean, you went to – you were the editor-in-chief of EDN. Yes. And then now you're back? You're back to EDA? I mean, why the jump back then?

Steve Liebson: It's a really twisted road, my friend. In 1996, a very good friend of mine came up with an idea for a new company, and I was going to be the director of technology. And unfortunately, within a month, we were at a business meeting with our primary client, and right in the middle of the business meeting, the founder of our company passed away.

Chris Gammell: Oh. Oh, my. Wow. Wow. That's terrible.

Steve Liebson: He just collapsed, cerebral hemorrhage, and he was gone within hours. Wow. My God. That was pretty much the end of the company. So – I can imagine. Yeah. The remaining partner and I hastily contacted a third friend who had been at Reed El Sevier, and we started a market research company specifically targeting high-tech companies that – like all the people that had been advertising on EDN magazine. That's called Beacon Technology Partners. It's probably, I think, still the market research company that's at the forefront of market research in the electronics industry. Oh, wow. And it's still being run by my good friend Jim McLeod Warwick. But after three years of that, I had this interesting alignment of planets. My daughter accepted a position to become a student at the San Francisco School of Ballet. And simultaneously, I found out that there was this position open at Reed El Sevier for somebody on the microprocessor report, and they'd moved me out to San Francisco, actually, Silicon Valley.

Chris Gammell: Wow.

Steve Liebson: So, you know, alignment of planets, say, okay, set that in gear, came out here, became a journalist again. Three or four people above me on the microprocessor report resigned within the first six months, and I was the last man standing. So I got to be the vice president of content, and I ran the microprocessor report and the microprocessor forum for a couple of years. Oh, wow. And then I made the jump to IP, intellectual property. I joined Tensilica, a microprocessor IP vendor that's sort of in the same market that our microprocessors are, but different. And then ended up joining another IP company called Denali for five weeks, after which Cadence bought Denali. Oh, wow. So that's how I ended up at Cadence. That is a heck of a journey, I have to say. You know, it's a Silicon Valley story, right? Yeah. Things move fast out here.

Chris Gammell: It sounds like a game of pinball, to be honest. I mean, where you're the pinball. Yes.

Steve Liebson: Yeah, that's, I've said badminton, but pinball will work.

Chris Gammell: Yeah. Wow.

Steve Liebson: Oh, boy.

Dave Jones: So you're now at Cadence. How long have you been there now?

Steve Liebson: A little over two years. Two years and five days.

Dave Jones: And we were just talking briefly before the show, and you mentioned that, well, explain what your job is there.

Steve Liebson: Yeah, that's a tough one. Well, my official title is marketing director. Yeah. But I write these two blogs for Cadence. And so one of the terms that I like the best is called brand journalism. And what that is, is I try and do the same thing I used to do on EDN, but I do it under the auspices of Cadence. And so the topics that I select are ones of particular interest to Cadence. But it doesn't mean that I write about Cadence all the time. It means I write about what's happening that's relevant to Cadence. And so in the EDA 360 Insider, I write about chip design and system design and developments that are related to that and interesting new chips that have come out and how they were architected and what's happening with deep submicron and nanometer design rules for creating chips and that sort of thing. And then for the Denali memory report, I write about everything about semiconductor memory. So DRAM and flash memory and the use of flash memory and SSDs. And I get a big kick out of writing about future, immediately future memory technologies, such as magnetic RAM and memoristers. Nice.

Dave Jones: So you were actually saying that also part of your job is to make public anything that you find out. So you deliberately try not to hear things within the company that you, otherwise you'd end up making them public and they shouldn't.

Steve Liebson: You know, that's sort of like need-to-know stuff, right? Right. If I don't know it, I can't possibly leak it. That's right. So I don't go seeking stuff that I shouldn't be writing about. And that doesn't mean I don't ever hear anything. But it certainly means that, for example, I don't bury my nose in company financials because I'm never to be a spokesman on company financials. And conveniently, if I don't know anything about the financials, then I can't say anything about the financials. All I can say is I don't know anything. Nice.

Dave Jones: That's a thing you find often in companies. Like don't tell such and such because then everyone will find out. So often there's certain people within the company who will stay out of the loop. You know, you'll deliberately keep them out of the loop. Yeah. Because you know that, you know, that person is the rumor spreader, you know.

Chris Gammell: Or don't tell a salesperson because they'll go out and sell it to me. If you have an idea for a product, don't tell a salesperson because they might go sell it.

Steve Liebson: Yeah. I think Cadence has a very sane attitude, which is don't tell anybody. Oh, boy. Don't make a judgment about whether or not this person may or may not leak it. Just don't tell them, period. You know, wait for a public announcement. For sure.

Chris Gammell: Yeah.

Dave Jones: Now, I've got to ask, with the CAD industry, like Cadence is one of the big players, right, which do everything from your chip level design right down to your, you know, your circuit board level stuff. Yes. But they make most of the money from their chip level stuff, right? Is that correct?

Steve Liebson: Yeah. Probably. The EDA seats are higher. They cost more. Higher margin. Yeah. But there's – Well, yeah. For IC design.

Dave Jones: Yeah.

Steve Liebson: And it's a lot harder to do that technology.

Dave Jones: Exactly. But, yeah. And that gets into my question is the number of seats have been – in those have been declining over time with fewer and fewer clients paying more and more. Is that like a sustainable model in the long term for the companies like Mentor and Cadence? Well, you know, that's sort of like –

Steve Liebson: Or is that a loaded question? No, no. Well, it's a Zen question because it's the same question that the industry itself asks over and over and over again, which is we're going to go to 20 nanometers. We're – you know, right now almost – the bulk of the stuff is done at 65 nanometers or 45 or 40 nanometers. Only a few companies have gotten down to 28 nanometer feature size at the moment. 20 nanometer is sort of cutting edge. Only a few designs have been done, mostly test designs. We're going to keep going. We're going to go to 20. We're going to go to 14. We're going to go down to 10. We think we can go down to 8. Madness.

Chris Gammell: Why stop there?

Steve Liebson: Keep going. Well, no. You know, pretty soon you say, well, now how do I cut this atom in half? Yeah. You can't – you know, then things blow up. You don't want to finish it. It's a bad time. You just want lithography, yeah. So, you know, so the question is how do the semiconductor vendors that are out there designing chips and the EDA vendors who are out there creating tools, how do they collaborate? And the foundries who are making the chips, and we've very much gone to a foundry model by and large, there are very few completely vertically integrated companies anymore, Intel being the prime example of a still vertically integrated device manufacturer. But most companies are either partially or wholly fabulous these days. How do these types of companies collaborate to keep Moore's Law on track? And so that's a kind of really interesting business question that the EDA industry is investigating right now. Because, you know, the question is right. The existing model of you buy tools from vendor A or vendor B or particularly vendor C, that's a good one to buy from, the vendor C one. Caden, why is that? No. No. Right. You know, it's – you can't really divorce the tool acquisition from going to the foundry anymore. It's becoming far more collaborative because the physics of what's happening at 28 and 20 nanometers is ferocious. And the amount of mathematical modeling that's taking place is mind-boggling at this point.

Chris Gammell: Right. And it's almost that the software itself is almost a whole part of the technology as well, right? Without the software, there would be no chips. Right.

Steve Liebson: If there are no tools, you know, it's just like saying, well, if there are no tools at Sears Craftsman, do we build anything? The answer is no. You don't build stuff without tools. And, you know, for designing chips, you don't build stuff without a whole lot of tools because it's very sophisticated these days. It's – and it was always that way, really. The EDA industry started when companies were internally developing their own tools. So in the 60s and the 70s, they were writing their own tools. So there was a lot of duplication of effort. There was not any – there was no standardization. But they got along. But those are the days when you could actually start an IC fab by going to the local photography store and getting an enlarger lens and some, you know, Kodak litho paper.

Chris Gammell: And some terrible chemicals.

Steve Liebson: And terrible chemicals and setting up shop with photo chemicals that you got at the local photography store. And, you know, we're pretty far away from that now. You can't do that. We are now at the point – well, we've been at the point for a long time. But, you know, at 20 nanometers, it's long past the time when you can treat light as photons. Right? We have this duality theory where sometimes light acts like a particle. A wave and a particle. Yeah. Pesky physics. Yeah. Well, at 20 nanometers, light never acts like a particle. There's only constructive and destructive interference of waves. And that's how we actually – Wave of probability. Yes. That's how we get imaging at this point.

Dave Jones: Does that affect the yield? Because there is actually that probability of it actually being there at that particular time rather than a point particle.

Steve Liebson: Well, we went quite a while ago. As soon as we went down below – most people don't know it, but the light that we currently use is 193 nanometers at the moment. Oh, I knew that. Oh, I knew that. Okay, good. So as soon as we dropped below 193 nanometers, we had to go to something called phase shift masks, which is where we started to treat light as waves instead of as particles. And so the math behind that is pretty good. The problems that we come into right now are things like edge roughness. So it's hard to draw a line that's got like a sharp edge to it. It kind of looks wavy, and that's because of the probability stuff. But the real yield problems are coming in just from having particulate contamination on the chip or – and this is getting bigger and bigger – variation of transistor parameters just from one place on the chip to another. Why on the wafer, right? Yeah, because we're using so few dopant ions these days that we used to be able to depend on dopant ions to work statistically because you could say, well, there's thousands of dopant ions in there. So statistically, they act like this. But when you have five dopant ions, the difference between five and six is not statistical anymore. It's significant.

Dave Jones: Why would anyone want to be involved in an industry like this? It sounds awful. Man, I can't imagine production on this stuff.

Steve Liebson: But you recognize immediately that there are people who love to tackle tough problems, right?

Dave Jones: Yeah, true.

Steve Liebson: We like to tackle tough problems, and what you define as tough may be different than what somebody who's doing 20 nanometer design tools is defining. But nevertheless, both of you have said, I can crack this tough problem, and by golly, I'm going to get a thrill out of it when I do.

Chris Gammell: That's correct.

Dave Jones: Speaking of this, you were recently at the Design Automation Conference, DAC. Yes, in San Francisco. But probably a lot of our, I was going to say viewers, listeners, sorry, don't know much about DAC. Can you fill us in, what it is, what happens there?

Steve Liebson: Sure. This was the 49th Design Automation Conference.

Dave Jones: Wow. Yes. It's yearly, I take it?

Steve Liebson: It's yearly. That's right.

Dave Jones: Is that the longest running technical conference?

Steve Liebson: It's supposed to be close to it, wouldn't it? Boy, it might well be. I mean, there were earlier ones, but they've long gone by the wayside. Right. It might well be. And as I said, you know, in the 60s and the 70s, companies were designing their own tools, and so they wanted to get together to exchange notes. And this was before there was an EDA industry, so there were no exhibitors. It was an academic conference. Right. It's an IEEE conference. And then over time, especially during the 80s, the EDA industry sort of grew up. And in fact, next year, which will be the 50th DAC, is also the 25th anniversary of Cadence. And so that kind of gives you an idea. Cadence kind of came in midway into the history of EDA tools. And so what you find there are you find the major EDA vendors, and you find vendors who are doing more niche-y sorts of things, but very interesting. And you also find the new startups who are trying to get some brand recognition by showing up and saying, hey, we exist. You need to know about us.

Dave Jones: And is that just one poor guy standing there in a suit with a tie in an empty, the lowest-cost booth?

Steve Liebson: Sometimes that is all that it is. Yep. But not all the time.

Dave Jones: No. I'm really surprised there's suit. They just have a chair. Oh, they have two chairs, a guy and a tie in an empty booth, and that's it.

Steve Liebson: Because that's all they could afford. But I've seen your videos of Australian trade shows, and that's the same thing.

Dave Jones: Right. Yep. We don't really have decent trade shows here.

Steve Liebson: Yeah, well, you know, there's a niche there.

Chris Gammell: Yeah.

Dave Jones: Yeah.

Chris Gammell: I'm really surprised there still are startups coming out. I mean, like, you'd think that, you know, some of our listeners probably follow venture capital type stuff. And the payback on this stuff is just terrible, right? I mean, like, it takes so long. The only way out these days is to have someone like Cadence or Mentor to actually buy them up. Or if there's a Fabless company, maybe they have, like, TI or some other big chip company buy them up. But I'm surprised that there are any startups at all these days. So what are these startups even doing?

Steve Liebson: Well, first of all, you need to understand that being a startup in EDA could mean as little as a couple of guys getting together with their laptops to write some code. So, you know, initially you don't need a whole lot of money necessarily. And interestingly enough, I was at the GSA, which is the Global Semiconductor Alliance. They had a spring meeting just a couple of weeks ago. And a very well-known investor in the EDA community, Jim Hogan, had a couple of gentlemen to talk. And one of them, his name was Dean Draco. And his company, he's the CEO of a startup EDA company called IC Manage. And he made the argument that, look, if you can create a $10 million business, which is about the size of a dry cleaning franchise, and you can manage it so it's throwing off 20% profit, that's $2 million a year. You can do that without venture capital money. And he said, personally, I can live on $2 million a year. So he was making the point that you could do a small startup with the intent of continuing to run the company. And he definitely has a contrarian opinion that you ought to start up a company to be a great company. And don't worry about the exit strategy. And if you are a great company, the exit strategy will come and find you at the right time.

Chris Gammell: Right. Very true. And if you're not worrying about investors, usually that's why you wouldn't worry about exit strategies as well, right? I mean, they're the ones that are usually pushing for that kind of thing. Engineers just want to build something. That's right.

Dave Jones: And if you are just two guys with your notebooks and your cutting code, you can live fairly frugally off that. Or you can even do it part-time. There are people who still have a day job and then go home at night and work on their startup. So you can work for years like that. And it doesn't really cost you anything except time.

Steve Liebson: No, it's certainly possible. I mean, there used to be a magazine published out of Rocky Ford, Colorado called The Midnight Engineer. And it was all about that.

Dave Jones: When did that? That was a long time ago, wasn't it? Yes, it was. The Midnight Engineer. That was 70s, was it? 80s, I think. 80s. It was the 80s. Okay. Yeah.

Steve Liebson: It was run by a guy who was, I believe, named William Gates, not the same as Bill Gates III. Right, right. And it published out of Rocky Ford, Colorado, previously only known for cantaloupes. Cantaloupes.

Dave Jones: Or as we like to call them, rockmelons.

Steve Liebson: Rockmelons. That's a good one. Rockmelons. Never heard that one.

Dave Jones: Yeah. There you go. That's it. Yeah. You don't come to Australia and, oh, I go to a shop and buy a cantaloupe. It doesn't happen.

Chris Gammell: I don't buy them anyways. Right. No. There you go. So at the DAC, I mean, are there a lot of big, like the chip vendors as well and the foundries, they're all there as well?

Steve Liebson: The foundries were certainly there. TSMC had a really big booth. The large EDA companies were all there. And about 200, I'd say, I'm just guessing, but roughly 200 smaller companies had booths all over the place.

Dave Jones: I have to ask, were Altium there?

Steve Liebson: I didn't see them. I don't think they were. I mean, DAC is primarily a chip EDA. Yeah. Sort of. Yeah, it's not really a PCB level. No, no, it's not PCB. And in fact, even the large vendors are not emphasizing their PCB stuff at DAC. Right. Because that's not the audience. Right. There are separate PCB shows where that's more appropriate.

Chris Gammell: Could you tell us a little bit more about that duality, too? Because, I mean, that's something that we deal with. We hear about DAC, right, and EDA companies. And, I mean, I always assume they're chip-level software programs. But those do exist within the same companies. You know, your company, Mentor also. So they have both of them. Is there a focus on one versus the other? Or is it really just kind of, well, they just kind of both are there and they do their own thing?

Steve Liebson: No, no. They're separate development groups, first of all. Oh, I agree with that, yeah. Because the code base is completely different. The problems being attacked are completely different. But there is synergy. So, for example, at Cadence, if you design a chip, eventually that chip's going to have to go into a package, right? It's going to have to be either bonded out or go through ball bonds to some sort of package that's been designed. And that package will generally have been designed with a PCB package, design package. And then that package is going to be soldered to a board. And that's going to be laid out with a board-level package, too. And you're going to want to be able to do simulation. Now, I mean, back in the day, you know, way before you went to school, Chris. Way before you were born, Chris. We didn't simulate traces, right? Because we didn't have to. Oh, my goodness. We didn't have to do the horror. Because they're 100 mil thick and they… Yeah, well, you know, and it was 10 megahertz. It was fast and 40 megahertz was the extreme limits of what you could do. But these days…

Chris Gammell: And you were clocking things. You were actually flipping switches up and down. Yeah, yeah, that's right. No oscillators. We used rocks. We used rocks.

Steve Liebson: You know, we just moved rocks on the board. Just tap them on the board. But these days, it's routine to have gigahertz signals running around the board. And we're routinely running 10 gigahertz, 28 gigahertz signals around the board.

Chris Gammell: Oh, yeah.

Steve Liebson: You have got to do electromagnetic simulation. And if you're doing that, you're going to want to be able to simulate it from the bond pad of the IC through whatever it goes through to the package, through the pins of the package or the balls of the package to the PC board, down the PC board trace, and then back up the stack on the other end. And so there is pure synergy there as far as electromagnetic simulation and the kind of information that you have to pass back and forth amongst these tools.

Dave Jones: I can imagine. Yeah. How would you have another tool, yet another tool that actually tied them all together and got out the one simulation result? Or how does that work? Because you have a common database is what you have. Right.

Steve Liebson: So 10 years ago, we just had the anniversary of this in a party. Cadence realized – well, before more than 10 years ago, Cadence realized that a common database was called for. And it created internally something called the Open Access Database and then turned it over to an organization called SI2 to standardize it. And now there are a couple of hundred companies that have adopted this Open Access Database for passing information back and forth amongst tools. And, of course, that's really what you need, right? That's the common format that you need to be able to handle that kind of thing.

Dave Jones: So you could have a startup. I'm sorry. I was just going to say – No, go ahead, Dave. Who's going to go? Should we do rock, paper, scissors?

Chris Gammell: Rock.

Dave Jones: Damn, you win. Okay.

Chris Gammell: Well, who's using these – who's using the Open Access? Could you give us an example of another company that might use it?

Dave Jones: Everyone, Chris. Hundreds.

Steve Liebson: Yeah. It's a couple of hundred EDA vendors. Okay. All right. So it's baked into the EDA tools from a couple of hundred different vendors.

Dave Jones: See, I didn't even know there were a couple of hundred EDA – Yeah, I didn't either. – vendors out there. I mean, it's just – and what – are you talking chip level here? You're talking as opposed to PCB level? Mostly it's chip level.

Steve Liebson: Yeah. But there is invaluable information to be mined at the system level as well.

Dave Jones: And that's where I was going to come in with my question before I lost the rock, paper, scissors shootout there. Or rock, paper, lizard Spock, which we'll get onto. Oh, Spock. Yeah. We can't leave without talking about that. You can't. That's my favorite story. Okay. We will end with that then. One of these startup companies, right, the two guys with their laptop, they could come in. Would it be common for them to come in and just tie two little obscure points of these databases together or something like that that nobody else has done before? Would that be a common startup kind of thing?

Steve Liebson: I don't think they would take that perspective. Usually it is somebody in the industry who is struggling with a problem and unable to find a tool to solve the problem. Yeah. That's what I was talking about. And they say, I can solve that problem. By golly, I'll start a company.

Chris Gammell: So in frustration, the industry just keeps throwing off startups, just based purely on frustration.

Steve Liebson: Well, you know, I studied this kind of in Colorado when I was there because storage technology was this huge storage company. And they would routinely throw off engineers, lay them off, which would start a whole new wave of startups within Colorado, which is where companies like Miniscribe would come from or the entire write-once-read-mini-optical-drive industry. And so the same thing happens in EDA here in Silicon Valley is that there is an ebb and flow of startups and consolidation and startups and consolidation because smart people, innovative people, creative people often get the itch and want to go off on their own and do something without other people telling them that they can't do it. And that's where you get startups coming from.

Dave Jones: Speaking of memory, you're the memory guru, right? You know all about that. You're one of the industry memory gurus. Well, that's what the Denali memory report says anyway. Exactly. What is the up-and-coming future for solar state memory? When will hard drives become, you know, a dodo?

Chris Gammell: Right.

Dave Jones: Well, not for a while. Right. And the reason is… By a while, are you talking 10-year time frame or… Possibly even longer. Well, that would be my guess too.

Steve Liebson: Well, the problem is that it's still the most effective cost per bit you can get.

Dave Jones: Yep. For storage.

Steve Liebson: And that's what it all comes down to, right? Comes down to cost per bit. And it's not just simple evolution. I mean, the disk drive industry has been able to pull rabbits out of the hat on a regular basis. Oh, yeah. So they come up with probabilistic coding methods for getting information off of a disk when they're not even sure whether that was a one or a zero.

Dave Jones: Once again, we're down into the physics again. It's crazy.

Speaker ?: Oh, yeah.

Steve Liebson: It always goes down to the physics. They come up with giant magnetic resistive heads that are actually small, but the physics is giant magnetic resistivity. Mm-hmm. And so they've done an excellent job of staying on their own version of the mortgage law curve, just challenging the semiconductor guys to come and follow them. And so they can't really excel in access time, but they can sure excel in cost per bit. And for a lot of applications, that's the number one figure of merit. Mm-hmm. That's the number one. That's the number one. That's the number one. It's still back in the milliseconds. So read is really, really fast. Write is really, really slow. So you have it.

Chris Gammell: That's, again, because of physics, right? Oh, yeah. You've got to pull things through the oxide. Oh, yeah.

Steve Liebson: You've got to make the electrons tunnel, and they don't want to do it, and it's probabilistic. I wouldn't want to do that. And now that we're down, I mean, Toshiba just announced a drive based on 19 nanometer flash memory, 19 nanometer. Oh, wow. So they're working. When they store a one on a nan flash cell, they're storing less than 100 electrons.

Dave Jones: Oh, that's ridiculous.

Steve Liebson: And, you know, from my dinosaur days, 100 electrons is like, are you sure? How do you know there's 100 electrons there? But they do. So we got two memory technologies that are vying to be the replacement technologies for nan flash. One of them is magnetic RAM, which is kind of amusing because prior to 1970, all machines— It was the thing, huh? It was magnetic cores.

Dave Jones: Exactly. All of the Apollo stuff, that was all magnetic memory. Oh, yeah.

Steve Liebson: The Apollo stuff was magnetic. You've been to the Computer History Museum. I can tell. Oh, yeah. Yeah. And then in 1970, Intel introduced the first commercially successful dynamic RAM, the 1103. And within two years, magnetic cores were dead as dodos. They were just gone.

Dave Jones: And that's practically overnight, right? That's practically—

Steve Liebson: Yeah, it's two years. After two decades of dominance, the 50s and the 60s, in two more years, gone. And that's how powerful a new memory technology can be if it meets people's needs better than what has come before. And unfortunately for magnetic cores, they did get smaller and smaller, but you could never automate the manufacture of a magnetic core plane. It was always woven mostly by women with tiny needles threading little bitty magnetic cores about the size of a pepper grain. Knit one, pearl two. Yeah, that's exactly right. Wow. That's crazy. And dynamic RAM just obliterated it. And so the magnetic RAM people and also the Memristor people are hoping to do the same thing with flash memory because both magnetic RAM and Memristor memory are non-volatile. They don't lose their bits when you turn off the power. And that's what it will take to replace NAND flash memory. Now, the NAND flash vendors are not going to stand by and let this happen. So they are exploring 3D versions of NAND flash memory, which are actually very interesting because they get away from the 100 electron problem by providing more interesting three-dimensional types of gates for the electrons to hang around in.

Chris Gammell: So it becomes like a cubbyhole system then at that point where it's like three-dimensional cubbyholes for all the electrons?

Steve Liebson: It's anybody's guess at the moment which of these technologies is actually going to win. But I think it would be an appropriate circle if the magnetic RAM were able to take back over from… That would be. Yeah, it would be something.

Chris Gammell: You know, it's annoying from the design side because you have to design based on the possibility. You know, I could design magnetic RAM. There's that stuff. There's ferromagnetic stuff available today, right? FRAM. Yeah. It's expensive, but… Well, and it's never… You know, who knows if it's going to be…

Steve Liebson: It's never gotten the densities. No. I've been writing about it since the 1980s. And it's been around for a long time. Yeah, I've been writing about it since the 1980s. Ramtron in Colorado makes… Right. …has been making those parts since the 80s. And you just can't get the densities up. It's a really cool technology. But, again, density is not there. However, that doesn't mean that it's not a good niche somewhere. For example, Texas Instruments just introduced a microcontroller family that's got FRAM instead of NAND Flash on the chip. That's right. You don't really have to worry so much about it as you might think because most of this stuff now is dealt with through a controller. So if you're talking to NAND Flash, you're talking through a NAND Flash controller. Cadence, for example, has an IP block that's a NAND Flash controller with some standard commands. And the controller is the thing that actually handles the protocols for the particular memory. So you could conceive of a controller that talked the same way as any other non-volatile memory controller on one end and talked MRAM or Memristor on the other end once those protocols got established. And it wouldn't really affect you as a system designer because you'd be talking to a compatible controller. And, of course, that's exactly what Solid State Drives did. They adopted the SATA interface. And the controller doesn't care whether the thing that it's talking to works by a spinning piece of magnetic media or NAND Flash. It really doesn't care. What it cares about is I'm sending these commands to read and write blocks of data, and I expect it to behave the way I expect. That's right.

Dave Jones: How much – when we're down in – we're talking 100 electrons, you know, to store a bit, right? You know, we're down in where the stats start to matter. How much of the chip and the chip design and the controller design goes into actually verifying and ensuring that the bit is written? Is there any actual extra circuitry to do that? Or is it just, well, we'll sort of, you know, cross our fingers and take our chances that the bit is in there?

Steve Liebson: How does that – Truthfully, I don't know the answer to that. But my guess is that there's little of both, right? If you're – Right. If your figure of merit is cost per bit, maybe you're just going to go on the statistical assertion. The stats, that's right. Yeah, the stats. Yeah. But if you're designing something that has got to be mil spec, maybe you check after you write, you know. Got it. Trust but verify.

Dave Jones: Uh-huh. Yeah, and that probably splits into the consumer domain and the, you know, aerospace domain, maybe, something like that.

Steve Liebson: Or enterprise class, even, you know.

Dave Jones: Oh, okay. Yes, yes, the enterprise class servers, which they use in, you know, the Google, you know, server farms that run our whole planet. Because, yeah, there's actually been some interesting reports. I think we've talked about it on the show before, Chris, where Google actually, you know, all of the cosmic rays hit and they, you know, flip bits here and there. And that is a big deal for them. Google have actually investigated that and they have to take mechanisms to prevent it and, you know, or at least help reduce it of these cosmic rays coming in and flipping these damn bits.

Steve Liebson: Right. So there you get into error correction codes. Error correction, yeah. Yeah, and there are increasingly advanced error correction codes that are coming out. Toshiba, they announced some new error correction code that I'd never even heard of before for this SSD that uses 19 nanometer flash.

Chris Gammell: Wow. Wow. Yeah. So you were talking about the hard drive manufacturers kind of doing whatever they can and the flash memory people doing whatever they can to kind of squeeze out Moore and Moore's law. Can you just tell us what the heck's, I mean, like, you know, we keep hearing, oh, well, they're running out of room, running out of room. And you even kind of got into a little bit before, but is there really a limit? I mean, what's going to really happen here, Steve? I mean, are we really out of room or is 3D? What's going on there?

Steve Liebson: Well, what's going on is that we continue to push the technology and it gets harder and harder to push. So eventually we're going to get to the point where it gets so hard to push that we're then going to adopt some other type of technology. One of the panels that I ran at DAC, we had Dr. Gary Patton from IBM and he runs their R&D group for semiconductor technology. And he gave semiconductor technology another two decades to run. So, you know, so we don't really know when we're going to hit the end, but we continue to investigate interesting new technologies that might someday lead to taking over for what we're doing today. And today we use CMOS, right? Everything's CMOS. But up till about 1980, everything was bipolar. Actually, until 19... That's right. You know, it was bipolar. We were going to go with bipolar forever because it got faster and faster and faster. And we didn't quite notice, but it also got hotter and hotter and hotter and hotter. And so we couldn't really stomach that anymore. And then we changed over to CMOS. But we haven't been doing CMOS forever. It used to be back in the day when RCA called it Cosmos instead of CMOS. That's right. Yeah. That stuff was so slow that you might as well... I know. You might as well be using relays instead of... I know. Relays or flipping the switches by hand. Yeah. And so if you told somebody that, oh, well, CMOS is going to be fast someday, they'd laugh at you. Well, CMOS did get really, really fast. And the companies that were counting on doing things in like gallium arsenide, they fell by the wayside because CMOS got fast enough. And so what do we got coming up? Well, graphene is coming up for one thing. We got this magic stuff called graphene, which is monoatomic carbon sheets. And it turns out that's a semiconductor, assuming you orient it the right way. And we're doing really interesting stuff with that. And it turns out there's a silicon form of it called silicine. And by golly, that can be a semiconductor. And we might be able to do something really interesting with that, but we don't know yet. And so we're going to continue pushing on all of this stuff until we can't push anymore because we really like Moore's Law. You know, Moore's Law is not really a law. It's a self-fulfilling prophecy. Well, it's a self-fulfilling prophecy. You know that if you do not invest in making Moore's Law happen, that the guy down the street or the guy across the ocean will. And so everybody continues on their merry way trying to make Moore's Law continue. And that's where innovation comes from. That's all these intelligent people who are around trying to figure out how to solve problems because they really like it. Well, there's plenty of problems to solve. And we will continue to solve problems. And we're going to continue to figure out how to get more transistors for the same amount of money because we continue to find new ways to use them.

Dave Jones: So in 20 years' time when Sagan grows up and enters the electronics industry, which he of course will, maybe not, he'll be using like a 7.4 GR series chip, 7.4 graphene series, will he?

Steve Liebson: Will he still be hanging in there? I think that's entirely possible. I mean, for all I know, they were right on Star Trek and they'll be using isolinear chips instead. So who knows what, you know, we don't know where electronics will be going. But if you were to ask somebody in 1945 whether we would stop using tubes or valves, they would say, well, no, there isn't anything else because the transistor would be invented two years later. And so if you were to ask somebody in the middle of the 50s if we would ever be able to build large computers with a million transistors, they would look at one of these things in a metal can and say, a million of these on one circuit platform? I don't think so. And then, you know, because the integrated circuit hadn't been invented yet, that was invented in 1959. And so we can't see these breakthroughs before they happen. We just don't see them. But they do happen. And then we figure out really interesting things to do with them.

Dave Jones: And they usually happen fairly quick, right?

Steve Liebson: Yes, they do.

Dave Jones: Are they happening quicker these days?

Steve Liebson: Well, there's...

Dave Jones: Do you think?

Steve Liebson: There's a thing called Liebson's Law, believe it or not. Oh, excellent. Tell us about Liebson's Law. I coined it at EDN. And it says that design engineers are conservative. And therefore, it takes 10 years for any new design technology to be widespread and adopted. And that's true for transistors. It was true for integrated circuits. It was true for embedded C. It was true for logic synthesis. It just seems that over the last 50 years, this time constant of 10 years has been relatively constant. Because, you know, we as engineers are often very conservative, especially for widespread application. Of course. You don't want to do anything too risky. Because the cost of failure is perhaps the termination of your company.

Chris Gammell: That's right. I was saying that before with the memory. Yeah. I was saying, oh, I don't want to design in new memory because it's not memory yet, right? Yes. I mean, it's the same thing. Yeah. So, like graphene, I don't want to design it in until I know it's there. But then it's kind of, how does it ever become designed in then, right? It's like, it kind of fits and starts, right?

Steve Liebson: It is. And, you know, the early adopters, they always have big problems. So, for example, it was really hard to learn how to use those early DRAMs. They were very finicky about timing. And you had to, you know, get the voltages just exactly right. But the same was actually true of magnetic cores. But after 20 years, everybody knew how to use them.

Dave Jones: Speaking of Star Trek. Yeah.

Chris Gammell: I was going to say, we got to get that. We got to hear about that. Okay.

Dave Jones: We are way over time. Well, we're over time. We're 11 minutes over time. You spoke about Star Trek. And you have an interesting story. I do. You were on a TV show with Leonard Nimoy. I was. Mr. Spock.

Steve Liebson: Mr. Spock. Tell us the whole story. Yes. So I actually, at the time, was the VP of content for a microprocessor report. And I was sitting in my office doing the usual things of the day, like drinking coffee and looking at my email. And into my office came a woman named Kathleen Dohler, who was the editorial director for a sister publication called Electronic Business. And she said, quote, some guy down in Burbank called me up and wondered if I would like to do a TV show with Leonard Nimoy. And I don't have the time. Would you like to do it?

Chris Gammell: Jaw drops.

Steve Liebson: Yes. Yeah. And I said, quote, let me think about that for a nanosecond, unquote. And I said, yes. So I got on the phone with these guys and they said, well, Leonard Nimoy was doing an infomercial series at that time called The Next Wave with Leonard Nimoy. And it would be broadcast at like two in the morning on CNBC.

Dave Jones: Why on earth would he do something like that? Money, my friend.

Steve Liebson: Money.

Dave Jones: Right. Okay. Yeah. He's got to make money.

Steve Liebson: Yeah, that's right. And so I guess the residuals weren't big enough on Star Trek. I don't think he's got that problem anymore because he doesn't do the show anymore. So I talked with the producers and they said, okay, so we're doing the show that we called about is on touchscreens. Are you an expert on touchscreens? And, you know, my internal response was, you know, dilithium crystals, matter, antimatter reactions, brain surgery. I'm an expert on anything you need me to be an expert on. I'm coming down. And I said, yes, I can do touchscreens. And they said, great. We're doing another show the same day in the afternoon on e-commerce. Are you an expert on e-commerce? And this was, you know, this was before Amazon got big. And, you know, this was before people were really just letting their credit cards fly over the Internet. And I said, well, yes, as a matter of fact, I am an expert on e-commerce. I'll do that show too.

Chris Gammell: So, sirs and madams, I challenge you to find something I'm not an expert in as long as the matter.

Steve Liebson: It didn't matter what they were asking. You're absolutely right. So, on the appointed day, I walked over, I drove over to the airport here in Silicon Valley, Mineta San Jose Airport. Took Southwest, which is, you know, not even assigned seats. You just get on the plane, it flies down there. And landed in Burbank. And then using MapQuest at the time, because Google Maps didn't exist yet. Using MapQuest, I discovered that it was only about six blocks to the Burbank studio. So, I walked from the airport to the studio and didn't meet a single person on the walk because nobody walks in L.A. as the song goes.

Dave Jones: Exactly.

Steve Liebson: And got to the studio and went through makeup. And the only time I saw Nimoy was on the set. But it was interesting because he was using his director skills at the time. And he was sort of running things from on the set where he would play a technical investigative reporter. He didn't really know about these technologies. That was my job to be brought on to the set and help him out. But, you know, he was giving instructions to the control booth. And we were intermixing the live stuff that we were doing with the stuff that had been prerecorded with a B unit at the companies that were sponsoring the infomercial. And we cut one of these things in the morning. And we broke for lunch. And Mr. Nimoy went off to his dressing room and didn't appear. And then when it was time for the next show, he reappeared on the set. And we cut another one. And then everybody said goodbye. And I walked back to the airport and flew back. And that was my day with Leonard Nimoy.

Dave Jones: Did you ever see it?

Chris Gammell: Did you ever see the final result? Or can we see it? That's a more important question. Can we see it?

Dave Jones: That's on YouTube.

Steve Liebson: I put it up on YouTube.

Dave Jones: Is it? Yeah. Oh, yeah. We'll definitely link that in. Yep. Oh, yeah. All right. That'll suddenly get an extra 5,000 views, I'm sure. That's brilliant. Yeah, it was a lot of fun. As if you would say no to something. Well, I guess some people would.

Steve Liebson: Well, Kathleen Doher said no.

Dave Jones: Okay. Right.

Steve Liebson: Because, believe it or not, there are one or two people in the world who are not Trekkies.

Speaker ?: Right. Right.

Steve Liebson: And somebody will now write in and correct it and say it should be Trekkers. Trekkers. Yes, I know. You've just started the war. To which I respond, tough. I was there in the 60s, folks. It was Trekkies. Right. And Chris is going, what the? Mm-hmm.

Dave Jones: What?

Steve Liebson: Well, what are these two old fellas talking about? I did three years of a comic strip called Star Trash based on Star Trek at the university. And again, Chris, it wasn't in your time, but I wrote for both school newspapers and my comic strip appeared in both of them, Star Trash. You can look that one up.

Dave Jones: Ah, so you're an artist as well.

Steve Liebson: Well, most of the comments kind of alluded to the fact that I was not.

Dave Jones: Ah, right.

Speaker ?: Okay.

Dave Jones: Could we see some of these early comics?

Steve Liebson: I doubt it. It would be hard for us to find. I guarantee you they're not on the web.

Dave Jones: Okay, so you didn't keep your original.

Steve Liebson: I did. They're not in good shape. I'm not sure they're scannable anymore.

Dave Jones: Okay.

Steve Liebson: You know, the newspaper has become extremely yellowed because they didn't exactly print school newspapers on archival stock.

Dave Jones: Yep. I was talking to Colin Mitchell at Talking Electronics. I interviewed him last year. The videos are up on YouTube. And he talks about how he used to lay out Talking Electronics magazine like a cow gum or something. So he used to stick down all these, you know, diagrams and everything all on these pages. Oh, yeah. So he put them under the photo typeset machine. Yeah. And he said all of his originals are now ruined. You know, it's 25 years later or something. And all the originals are ruined because all this gum stuff, all this glue used to stick them down just turned, rotted all the paper away and turned it yellow and you just can't read it anymore.

Steve Liebson: Yeah, that's exactly what happens. It's all ruined. Yeah, we weren't shooting for that stuff to be archival, right? No. It just had to be fast. Yeah, that's right. Had to do the job.

Dave Jones: Oh, well over time here. But that was, thank you very much. My pleasure.

Steve Liebson: It's been a lot of fun.

Dave Jones: Oh, any last minute questions? Have we lost Chris?

Chris Gammell: I think he fell asleep.

Dave Jones: He fell asleep. He's too old, fellas.

Chris Gammell: Oh, no, no, no. Sorry, I'm here.

Dave Jones: He's back. All right.

Chris Gammell: Yeah. No, I was going to tell people that we should tell them where to find Steve. So eda360insider.wordpress.com. I highly, highly recommend people subscribe to the feed. It's an awesome, awesome look inside, you know, all the EDA industry, chip industry, fat. What's happening? Yeah.

Dave Jones: Edding edge type stuff, usually.

Chris Gammell: Yeah. Right.

Dave Jones: Stuff you don't usually get to find out about.

Chris Gammell: Exactly. Yeah. Access to, I remember the one interview you had with one of the VPs at TI talking about analog technology. I love that interview. And just tons of great stuff on there. So I highly suggest...

Dave Jones: And you've got a personal blog as well? Do you? Well, no.

Steve Liebson: The other blog that I do is denalimemoryreport.com. Right. Denali is the mountain in Alaska. Denalimemoryreport.com. And then I have a personal history site on the HP 9825 computer, which is conveniently called hp9825.com.

Chris Gammell: All right. Easy to find. Easy to read. Check it out, folks. Yeah. Well, Steve, thank you again so much for being on the show. We really appreciate your insight. We look forward to hopefully talking to you again in the future.

Steve Liebson: Anytime you guys want me back, happy to come back. Great. Terrific.

Chris Gammell: We'll talk to you then.

Steve Liebson: Catch everyone next week.

Chris Gammell: Okay.

Steve Liebson: Thanks. Thanks.

Steve Liebson: We'll see you next time.