If you’re involved with IP and/or EDA, you’re probably already acquainted with Ian Mackintosh, Victor Berman, Ralph von Vignau, Bill Martin, or Warren Savage, but don’t let these men fool you. They may be affable enough and a load of laughs in the lobby or the lounge when you see them around the conference circuit, but they’re deadly serious when it comes to the business and game of standards, particularly IP standards.
They’ve got skin in the game, they’ve left skin in the game, and some of them have been skinned alive while playing in the game. But win or lose, these Tough Guys bring their A Game to the Rough Game of standards, and when they’re on the court, they're definitely in the paint. (Yeah, okay – maybe a tad too much NCAA basketball of late, but you get the point.) These guys have put thousands of hours into the struggle for standards domination with a list of involvements that reads like a roadmap of the last several decades of play: Accellera, FSA, GSA, IEEE, OCP-IP, OSCI, Si2, SPIRIT, and VSIA, to name just a few.
This discussion took place Last Month at DVCon in Silicon Valley, and was prompted by a panel Last Year at IP’07 in Grenoble. However, there’s nothing Last Month or Last Year about what’s covered here, because the trends these men are commenting on will have ramifications Next Month and Next Year, and for a long time after that. It was a great pleasure to listen to Ian, Victor, Ralph, Bill, and Warren en masse. Thanks to all of them for their time, and to Linda Marchant of Cayenne Communications for arranging everything.
[Before going on, do two things: 1) get that proverbial cup of coffee, and 2) click on that “print article” icon up on the right, so you can see this text all on one page.]
Peggy Aycinena: Gentlemen, if you will – the short, impromptu version of your bio.
Ralph von Vignau: Philips Semiconductor, NXP as a spin-off, EBU, CEPT, CCITT, ISO, and lately SPIRIT and OSCI.
Ian Mackintosh: OCP-IP, Mentor, PMC-Sierra, VLSI Technology, now NXP, National Semiconductor and several start-ups, board of VSIA, chair of the VSIA IP Working Group there, and VCX, as well.
Warren Savage: Fairchild, Tandem Computers, Synopsys developing DesignWare in the early days, IPextreme. At Tandem, worked heavily on SCSI standards, fiber channel, and saw enough in that process to hope standards could be somebody else’s job.
Bill Marin: Mostek, VLSI Technology, lots of customers, taped out 60 or 70 designs of my own, Synopsys as part of the physical design team there, Mentor Graphics, VSIA, GSA.
[Editor’s note: scan to the bottom of this article for additional info on these gentlemen, and additional info on the various standards organizations referenced here.]
Peggy Aycinena: Let’s start with a simple question. Why do we need standards?
Victor Berman: We need standards to enable the efficient growth of the industry. The whole point is to make the routine aspects of design both routine and automated, and not waste people’s time. Standards allow tools to be interoperable and allow people to quickly ascertain whether a piece of IP is what the project needs. For IP users these are the basics.
|Ralph von Vignau|
At IP07 last December in Grenoble, I talked about the need for these standards, not just to make things interoperable, but also to [be able to] rely on better systems. We need the same set of protocols to develop testing procedures that are standard across the company. We need an ecosystem of players working together, so the end users can have confidence [in third party IP]. That’s particularly important in certain safety critical domains like automotive. The IEEE is in the same boat. They’re using a lot of time getting quality around standards. Standards are important because they lower the cost [of IP] and level the playing field across the market.
Ian Mackintosh: The existence and need for standards is a fundamental belief and critical for sharing efficiencies. Standards grow markets and enable people to mix and match the best of breed in IP. Standards define the level of maturity in an industry, because all industries eventually adopt standards. It’s probably a sign of the lack of maturity in SoC development today that there are not, as yet, a broad range of standards there.
Ralph von Vignau: I’d go one step farther and say the world almost wouldn’t work without standards. Planes, trains, your TV, your phone – they’re all built up on standards. [Down to the] lowest depths of an IC, everything has to conform to standards. Baseband, 3G, they’re all based on standards. There are a lot of standards for systems and functions in the electronics industry, but there are few standards that regulate actual electronics development at a base line. Everyone is fighting each other on this, but we’re starting to learn that building market-based standards is what is needed today.
Ralph said it best: You’ve got to tie into these standards. Whether it’s USB or a file format like GDSII or OASIS, standards have helped people to be a lot more efficient. Go into any restaurant, and they’ll give you a menu to of what they can make in their kitchen, so they can order their supplies in advance. Standards are everywhere. Some are very rigid like USB, and others are more localized like that restaurant. But either way, we’ve got to have standards.
Peggy Aycinena: I’m impressed with all of the noble sentiments you gentlemen have expressed here. But in truth, isn’t there really a lot of resistance to standards?
Victor Berman: Frequently, people want to keep technology proprietary and want to keep a lock on the market, whether for right or wrong. They think it’s better to keep things proprietary rather than sharing. Most of us here in this room, however, feel that there’s limited capacity [in that model] for growth.
For a market to grow, you have to have confidence that you can produce a better product based on differentiation rather than proprietary technology. Large companies with dominant market positions may be less interested in standards, because their way of doing things is the de-facto standard. [Also, some people think] to develop a true standard takes a lot of time, the technology will be watered down, or they will lose their competitive advantage.
Then there are the perceived costs [associated] with developing standards – the time element, the physical activity in terms of engineering resources plus the money. It’s expensive to implement standards because you have to change your technology to a certain extent. And, when it’s not clear that if there’s a single standard, as mentioned earlier, the equation is even more complicated. You have to guess which standard to support, or if there’s more than one standard, should you use your own internal formats and then output what you want as a translation? So, yes – there are good and bad reasons for not having standards.
Also, standards force you to deal with your competitors, to find a way to work and play well together. That’s not always easy. It’s quite complicated, particularly in some of the examples cited here. People had to give up something to work with USB, for instance. But in EDA, it’s not just function and form factor, it’s also methodology, which is even more controversial. Agreeing on particular languages or formats is very difficult. And in IP, there are lots of things that go well beyond just making things work together – documentation, efficient transport, and so on.
Warren Savage: This whole thing got started from a blog I wrote when the VSIA initiative folded. One of the tough things about IP standards: they’re not easy and it takes a lot of effort to keep things going. It’s not like SCSI or USB, where there’s an obvious evolution. As VSIA and [other such] things in the past prove, it’s hard to keep these things alive. The investment costs for IP standards are continuous and, in the case of IP in particular, it’s often big companies trying to block out small companies. [Those companies believe] standards thwart innovation, but if you define your negative space that way, it increases your positive space.
Peggy Aycinena: That sounds very New Age. [Laughter]
Warren Savage: There’s one more negative thing that might be perceived about standards. They tend to force commoditization on pricing, which often drives prices down.
Ian Mackintosh: I disagree. There is no intrinsic value in standards, but there’s infinite value to be leveraged from having standards. That’s why we have trouble with people being opposed to standards. It’s the NIH [Not Invented Here] issue, the parochial issue, and the lack-of-vision issue in terms of understanding the opportunities. [We need] education plus vision [in standards work]. There’s also time-to-market pressure, which turns a lot of people off [as far as standards are concerned]. Plus, there’s a user-focus issue, because some people need things from standards that are very different from what others need. All of this together causes conflict and fragmentation, and a fundamental fear of the complexity of issues around standards.
Ralph von Vignau: I’m a bit of a devil’s advocate here. In principle, we could live without standards. [People can say] standards stifle innovation and development, they keep people in a herd rather than keeping them imaginative. [Or they say], a level playing field? Well, I want to make an impression on the world, but that level playing field destroys individuality and destroys my drive to do new things.
Start-ups are [all about] breaking out of the world of standards. So, could we live without standards? It’s surprising, but sometimes people do. For instance, there is a railway that goes throughout Europe. It’s narrow gage on one side of a particular border, and wide gage on the other side. Obviously, however, there are certain safety factors where standards make sense. But I often wonder – and don’t get me wrong here, because I believe all of us here do good work – what would the electronics world look like if we didn’t have standards. We’d have 10 different kinds of phones. I don’t [have an answer to this dilemma], but humans are creative, yet standards are not always creative in what they cause people to have to do.
Peggy Aycinena: For the record, we’re only permitting Ralph to utter this type of heresy because he is the President of SPIRIT. Otherwise, he’d be voted off the island. [Laughter]
Bill Martin: It’s true that standards do stifle creativity in people trying new solutions. But in business, can you always make money if you do the proprietary solution? There’s often a lot more cost [associated with that route]. Apple’s a great company, as an example, and there will be a small, strong population of MAC users forever. But you pay the price if you’re one of them. It’s a more expensive computer [than a PC] and there’s a lot less people buying the software, so you pay an inherent cost for that individuality.
There’s nothing to prevent product companies from going off against a particular standard, but the overall cost of creating a market and making it a profitable venture [has to be considered]. You have to ask – do we start by going down a brand new path, or do we want to tie into existing standards and build a business out of that? The first way is very expensive – yes, there’s creativity, but if you look at the market pressures, and the dollars involved in rolling out a product and getting a following, there’s a very high price to pay for going the proprietary route.
Peggy Aycinena: Well, Apple may be a great company and their MAC and iPhone are all well and good, but I’m offended by what I see as a major lack of standards in the cell phone industry. The iPhone’s a lock-in for AT&T, and every provider requies a unique piece of hardware with no interoperability between different service providers and handsets.
Victor Berman: That’s mainly the case here in the U.S. It’s actually better in Europe, and China is wide open. And all of us here, despite our complaining, are actually guilty of being enablers of the cell phone companies in this area. They all offer these enticing products [with the phones]. If you’re strong, of course, you do something else, but it’s hard to resist the marketing, so we sell our souls. Apple has sold more than a million phones connected to AT&T, [and now] people are cracking the iPhone code.
Bill Martin: Yes, now people are even posting information out on the Internet describing how to deconstruct the iPhone and open it up to any service provider.
Warren Savage: That brings us back to what Bill said earlier. There’s definitely money to be made by not being forced to adhere to standards.
Peggy Aycinena: But I thought the money was made by adhering to standards?
Warren Savage: Actually there’s money to be made on both sides. It’s actually two sides of the same thing.
Victor Berman: You can have two different GSM phones, but they’re both locked because they have two different protocols.
Ian Mackintosh: I don’t agree. I don’t think there’s a lack of standards in cell phones. Look, cell phones do talk with each other across service providers, so standards do exist in that industry. The real question is – at what level do they exist?
We live in a capitalistic world, where companies exist to make profits. Protecting channels, or incomplete standards, or interoperability is, thank goodness, a choice. We should leverage those opportunities while they exist. It’s only through a natural evolution that handsets will be interoperable, which is [by the way] definitely what’s in the cards today.
Ralph von Vignau: This goes back to our basic theme. I also believe that it’s a choice that everyone has. Codes never last for long. Look at the DVD [anti-piracy code] – even it was broken. There’s a battle between the cell phone service providers and the equipment providers. Service providers lock you in for 2 or 3 years, because they’ll get back [the cost of the handset they’re providing for free] through the calls. In certain European companies that’s been broken up by legislation, not because the service providers did it, but because legislation stopped it. In the U.S., the question is who or when will the throttle on the system be broken. [Until then], the service providers will retain the power to throttle change.
Bill Martin: A lot of good things are being said here. Ian said cell phones work between each other. Yes, that’s enabled by standards. And, business models provide the differentiation, so there can be a lock-in. I do think in the next 3-to-5 years, you’ll have what you want, Peggy, as far as choosing providers wherever you’re at.
Things like the open operating system initiative from Google and Verizon is going to break down some barriers [and help answer the question of] who owns the content in providing additional software features in the phone. It may come from different providers in the future, but this [controversy] is going to continue. In time, however, you’ll get what you want. Some competitors are now working in tandem. As soon as you see that happen [in an industry], you know change is coming.
Peggy Aycinena: So, speaking of change. What happened with VSIA? Can you guys talk about it?
Victor Berman: VSIA lost its focus. They started out with a very different idea of what they eventually were trying to do. They largely accomplished the first piece [of their intended agenda], and then the organization became quite large, but wasn’t focused enough on specific goals. Initially, there were similar goals [across the membership], but that evaporated over time, so it was hard to keep things going.
Warren Savage: I’ve been tracking VSIA for 10 years, and I [would argue the focus was lost] earlier more than later. They overreached somewhat in what they were trying to do. It was originally a big and noble idea that, had it come to fruition, would have been good for the industry. But they always battled uneven participation, or lack of participation. Some VSIA working groups were very sparsely attended, and for those groups it was hard to even get a quorum. Then, with time, I saw the bigger companies lose interest. They started to ask, what’s the ROI of participation in VSIA?
Ian Mackintosh: All standards organizations suffer to one degree or another from a set of problems. For VSIA, the primary problem was their inability to communicate the value, the benefits, and the facts [about the organization]. It caused a failure to get by.
However, to a different degree – they improved the ease of adoption for the things they were working on, and the actual value in terms of standards themselves. But the timing of some of the things they were doing [generated FUD] from other standards bodies – issues around NIH, protectionism, were standards compulsory or mandatory. [Finally], they faced the issue that I call adoptive diversity. People always have different concerns and needs that [they hope will be met by adopting] standards. That diversity can pull apart the activities [in a standard] and cause doubt about the ROI of participation.
Ralph von Vignau: Certainly, there was some de-focusing at VSIA. They forgot who they were standardizing for. Their efforts became very inward driven, rather than looking around at what they needed to provide. They were only working for themselves in the end. It brought to an end the innovation and the forward drive, and VSIA disappeared. The working groups collapsed. When people gathered [at VSIA meetings], they were talking about esoteric problems of the world. That was the worst of it.
In the end, it became a discussion of the organization and the members achieving their own goals as people, rather than what are the standards. Of the four standards VSIA was working on, if it hadn’t been for Kathy Warner driving QIP, there would have been nothing accomplished at all.
VSIA was a learning process for other standardization groups – people watching to ensure it didn’t happen again. In a way, we’re all part of [the history of VSIA]. I was even on the board of VSIA out of frustration, but you just couldn’t turn this thing around.
Bill Martin: I was in late [at VSIA] – Ralph was already gone by that time. Victor was there, but then off a year, and Ian had been in VSIA before I was there. I was there at the tail end. Early on, I could see it was way too late. There were political fights that caused a lack of things getting out early, and I agree – there wasn’t any focus.
Without results, you get a bad reputation, so people started dropping out although there were still members. But, there was apathy. People wouldn’t sponsor different working groups. I think the early 3 or 4 years of the organization set it up for failure. It was just a question of when. I was one of the ones who voted to shut it down.
But, it’s very important to note that Kathy Werner did not cause the demise of VSIA. She should not take any blame at all. Kathy put her heart and soul into getting the Verification and Hard IP QIP in place. The sole credit is hers for getting those things done!
Warren Savage, Ian Mackintosh, Ralph von Vignau: [Together] Absolutely! Kathy should take no blame, but only credit for what was accomplished by VSIA.
Bill Martin: In fact, if we’d had a ton of clones of Kathy, VSIA would undoubtedly have been very successful. It would have been much more like what SPIRIT is trying to do today.
Ralph von Vignau: We were actually in the process of attempting to merge the VSIA and SPIRIT consortiums, but it just could not be done. This organizational beast kept looking at itself like a sort of self-obsessed entity. Both Gary Delp and Kathy worked with me on this, but [it could not be done].
Bill Martin: So, Kathy should not blame herself at all. She was the only one who was driving things, by the end.
Ralph von Vignau: Especially after some of us left VSIA. But we felt if we couldn't improve it, we needed to leave it.
Peggy Aycinena: Well, then should we blame Ralph? [Laughter]
Bill Martin: Not at all. Again, VSIA was set up a long time ago for failure. Even 5-to-7 years ago, it was already on the railroad track to disaster. Kathy managed to land it safely and without injury to people. She should be credited completely for that.
Victor Berman: Kathy did the right thing, because like a lot of groups, VSIA could have continued on forever.
Bill Martin: Victor, you were there when we said, let’s get Kathy in there and see [if it can be fixed]. But after you left, Stan Krolikowski and I talked about the possibilities. VSIA was a non-profit, but could still cover its expenses. They were still bringing in new members and all that, but it was a very tough go and a tough decision.
Peggy Aycinena: So, moving on. What’s great and good in standards today?
Ralph von Vignau: SPIRIT, for one. The main reason for SPIRIT, which we’ve made very clear, is that if people want to contribute, they have to give money, their presence, and engineering. That’s the major difference between SPIRIT and VSIA, for right or wrong. Also, we’ve limited structurally the number of companies who can partake in working groups. That allows the groups to be more effective.
Things are made very clear at SPIRIT – it’s a very clear project-driven approach. You have a job to do, a timeframe within which to do it, and you need to get it done within that time. Most people are much happier this way because they feel the ROI and their time are respected, and because it gets results. These are the real differentiators at SPIRIT, and a lot of it comes from what we learned from the VSIA experience.
Ian Mackintosh: So, I would focus on OCP-IP to see what’s great and good today. A lot of things were learned from what didn’t work at VSIA. That’s the good news. That experience has spawned better thinking about how things are done, and I can see the results around this table today. In OCP-IP, for instance, we’ve done very well on infrastructure, development, incorporating contributions from the working groups and corporate members, and from collaborations with other associations and universities.
As far as time to market [with our standards], we’ve really done well there as well, driven by the strategies of need and greed. We don’t do anything as an organization unless somebody wants it and somebody else is working on it. We’ve got flexibility built into our operations and our relationships. And – none of those relationships are standard!
Peggy Aycinena: So there are no standards in relationships? [Laughter]
Ian Mackintosh: No, we find the work gets us across that interface, [particularly] as technical excellence has always been our primary goal. And, we have also always cared about our website communication – we’ve always wanted that to be [an integral part] of our infrastructure, and it has worked for us.
Warren Savage: I’ll start with Ralph and Ian as being great and good things in standards. These two guys are driving some of the most successful standards today, and they’re providing a showcase for how to do it. And I agree with Ralph, these successes are built on various trails at VSIA. It’s from their involvement there, and what they learned there, that they’ve been able to push things forward.
Of course, there are some de-facto industry standards in place today, the most famous being the ARM AMBA bus, but it’s not a consortium putting it together. It was put together by one company, but once out in the market it’s been a useful standard for all to use. ARM still retains ownership of the standard, however, yet that seems successful in the market.
In my sphere of work, I’m working with Infineon on some automotive standards, trying out propositions on the market in a similar way. [The lesson is], there is room in the market for some proprietary standards being put out into the public sphere, but there’s also room for collaborative standards.
Ralph von Vignau: I would challenge the AMBA standard [analogy]. I see a marked difference between OCP and AMBA. If ARM didn’t have almost a monopoly with everything including mobile phones, would they have had the success with their standard?
Warren Savage: Everything is related to an ecosystem, as you’re aware from Phillips. Remember the PI Bus, which was European and got only so far. ARM took the ecosystem and was an early IP player. If you were a processor [provider], it was a natural place for the standard to grow. It was almost the same way with Intel and the PCI Bus. Before, there were lots of busses with no standardization and it was very expensive until Intel came out with their standard.
Bill Martin: Yes, and it [preceded] a huge sale of Intel processors. It was the same thing with ARM. The end goal was [dominance] in the market.
Ian Mackintosh: I see there are basically two ways that standards evolve. I call it “The Rule of the Most Fit.” Absent of competition, the PCI standard or GDS II are intriguing. I call those “Force Feeding.” I see SPIRIT, Si2, and OCP-IP building the two types.
Victor Berman: The AMBA Bus is a good example of where there was a proprietary standard, and there was a vision plus selfish purposes to open it up, but what becomes the repository for maintaining it? Similarly with VSIA, they had ideas in a particular area, but become diverse, unfocused and hard to manage. So a number of organizations have sprung up, but what happens next?
This is where IEEE comes in. It has several unique and beneficial characteristics. It has name-brand recognition. It has processes that are well understood and continually evolving. It deals with issues that are real in the industry – IP rights, how to promote technology – and IEEE has been very effective.
One of the reasons it’s effective is because IEEE is not a monopoly. At VSIA, there was one small board trying to manage different technologies, not all of which they understood. [On the other hand], the governing bodies of IEEE do not try to meddle in the technology. They specifically try to have working groups in particular areas where the expertise lies. As they evolve or shrink, IEEE is perfectly happy for working groups to disappear, even as the parent organization remains in place.
In VSIA, where did you find the documentation? And [there was always the question], was the organization going to disappear? We’ve never had that issue with the IEEE. There’s something to be said for individual, focused organizations to become feeder organization for IEEE, [particularly] at a certain point of maturity. An organization can hand off their stuff that way, and have it formalized and made open.
There are some of these characteristics in SPIRIT. They’re effective by being narrow. That has always been a reasoned and widely discussed aspect of the organization, but when you get to a certain point where you want broader adoption, what then? There’s a good reason for an innovative organization that can do marketing and can reach out to companies to understand their pain points. But, you get into trouble when those organizations want to remain [in place] forever.
Ralph von Vignau: Also, one of the things we learned from VSIA – from the start at SPIRIT, we didn’t produce the standard, but fed into the IEEE. VSIA tried to make its own standards, but IEEE is recognized and has as a certain feeling, even in certain countries. A small standard group can’t produce a global standard.
Victor Berman: Sometimes it is better to remain small.
Bill Martin: I think Ralph really did learn from the VSIA – let’s donate this into the larger group. And even though it was an older group, it did learn new tricks through the passion of the people there. I hope that reinforces the work at Accellera, and I hope that you guys at SPIRIT will donate [technology to the larger group].
Let’s get the train moving this way, because once things become standards, [you’ve got to take] the ownership away. Being a Show-me kind of guy, for me the jury is still out until these standards get out and are really adopted by suppliers, and consumers are really using them. A lot of good work is going on, but only time will tell if any of these things are really beneficial. For the various tools and IP vendors – for anyone who supplies something along the chain – they need to start to support standards. The consumers demand standards.
Peggy Aycinena: Does that really make sense? Do consumers really know what IP standards there are, or where they’re used?
Bill Martin: Perhaps not with respect to IEEE or USB standards, but you can do a Plug Fest and get thumbs up or down from designers. Those are easy things to go off and certify. But, if you get into how the IP was developed and documented, and the kinds of tools that were used to do that work correctly – right now, everybody’s got religious wars going on because everybody’s got their own methodology.
That was the one major problem I had with the VSIA QIP. What you really need is a spider that goes through and looks through all the documents to see what kinds of tools were used to develop IP, and what documentation exists. You’re never going to have a perfect solution, but until you have more documentation, [there will be problems]. Using IP is built more on relationships, it’s based on who do you trust to deliver quality IP.
Peggy Aycinena: What are the hurdles involved here. And more importantly, what ticks you guys off the most in all of this?
Victor Berman: There are a number of things. Funding is one. Whether to fund this standard or that standard or not. Coming up with a roadmap is another. It’s easy for me to come up with a top-down roadmap, but to be effective, you have to address the pain points of the big companies, or you’ll get in the way of some initiative. And, that can be very difficult. Also, a lot of successful standards have been ad-hoc. No standards ever come along because some great planner plans it. Standards only come along because of some discontinuity in the industry.
You ask, how do you know if people are using standards? The answer is, you don’t! Verilog is the most interesting case. Somebody buying a cell phone has no interest in whether it was designed using Verilog or not, but you can be sure the phone was developed with a standard HDL. That’s an aspect of standardization that everybody understands and can see the value in. However, that wasn’t true until recent times – HDLs themselves are relatively new. They all used to be proprietary.
In IP, which is new, things are much more difficult. SPIRIT is addressing a few of the aspects involved, but they’re only scratching the surface. There are still low hanging fruit [types of problems] in IP that are not being addressed because the resources just aren’t there. And, the burning desire isn’t there. So, it’s pretty much a free-for-all right now and definitely a problem.
There’s no overall organization that shows people the way, that educates people, that’s trying to put them on a path to improve the industry. I’m attempting to hook up with GSA to promote more regularity, but I’m not sure how that will work.
The other problem is that standards tend to be U.S.-centric or occasionally Euro-centric. The emerging countries with the largest markets and largest numbers of developers are excluded by language and time zone. It’s not a conspiracy, but that is the end result of the situation.
People in countries like Japan, India and China feel very excluded from what’s going on in standards. Things are being driven at them without any input from them. Ironically, VSIA had just started to penetrate into China, just as the origination was falling apart. VSIA was one of the few standards groups recognized in China. There needs to be more outreach to these countries. I’m working a lot now in China, trying to encourage people there to put mechanisms in place to address this big problem.
I think the globalization aspect of standards is very important. In SPIRIT, we made several attempts [in this area], and have been marginally successful, but it’s difficult and expensive. There are lots of language problems, the distances are great, and you need somebody with a burning desire or business interest to pursue these things. There are still people trying to railroad their proprietary stuff out as standards, so there are multiple standards emerging and people are starting to bump heads. It’s just human nature, although I have seen some cooperation [lately] between Mentor Graphics and Cadence.
Peggy Aycinena: I call it the Venn diagram of overlapping initiatives. It’s Cadence and Mentor in verification [OVM], versus Synopsys [VMM], while it’s Mentor and Synopsys [LPF] versus Cadence [CPF] in low power.
Warren Savage: In other words, why can’t we all get along?
Victor Berman: I’m sorry not to be more nasty about all of it, but compared to what it was like when I first entered EDA, it’s still so much better today there’s no comparison.
Peggy Aycinena: Warren, you’re next. What ticks you off in the industry? And, what hurdles remain?
Warren Savage: First of all, I concur with Bill. My chief complaint was the VSIA QIP. I go back to my days at Synopsys, when evaluating IP was highly qualitative. This concept of a spider crawling through the deliverables [is a good one].
Bill Martin: It’s [a case of a] subjective versus an objective evaluation.
Warren Savage: You can’t have the marketing guys filling out the QIP, which is what quite a few companies were doing with the VSIA QIP. That doesn’t help the standards effort at all.
Bill Martin: Are you telling me, Warren, that you don’t trust me to fill out the QIP for my IP. I can’t believe that! [Laughter]
Warren Savage: I also agree that part of the standards thing is about the big companies seeing all of this as market orchestration. I’m cutting to the cynical part of my response, but unless you can get these big companies to think about why standards are not just good for them, but good for everyone, you’re not going to make progress with standards. We need to make them comfortable with these ideas if we want to make progress.
My biggest gripe? It’s with SystemVerilog. It was a noble effort to develop a next-generation language that could supplant VHDL, Verilog, and other proprietary languages, but it turned out to be a rush to standardization so people could claim a new standard language.
But, as I’ve said in the past, so far it’s inappropriate as a design language because even though the EDA vendors all signed off on the spec, they have been very inconsistent with tools to support it. It’s a Swiss Cheese market situation. You can’t be sure that a SystemVerilog-designed piece of IP will be supported by Mentor, Cadence, and Synopsys at the same time for your customer. It’s all a very big disappointment in my opinion.
Ralph von Vignau: Was SystemVerilog a panic reaction against SystemC?
Warren Savage: I don’t know.
Ralph von Vignau: I’ve often had that impression.
Victor Berman: It was more a reaction to e, than to SystemC.
Ralph von Vignau: But e is not really a system language.
Warren Savage: There were people putting out free e simulators, now OSCI is putting a free SystemC simulator out on the market.
Peggy Aycinena: Ralph? Your biggest gripe?
Ralph von Vignau: The EDA vendors! [Laughter]
I think one of the most damaging affects we have today is from having 3 very big EDA vendors – big, that is, in comparison to the other EDA companies – who have stifled in the past and continue to stifle innovation. You have to kick them to get anything out of them! If it wasn’t for the small vendors providing solutions piece by piece, there would be no progress. You can’t have a standard without the appropriate tools to support the standards, so standards have no value in the electronics industry today.
We saw how hard it was to get standards out of the EDA vendors, but they simply won’t do it until they see the customers supporting the tools, but there are no tools – so it’s clearly a chicken-and-egg situation.
Ian Mackintosh: We have actually figured a way around this. The small EDA vendors are very amenable to growing the market. They are the small, nimble guys who support standards development.
Ralph von Vignau: So without the small companies there would be no progress?
Ian Mackintosh: The support comes, as always, from the motivation of need and greed and there is more of that in the smaller companies.
Ralph von Vignau: So, everytime – it’s the little ones pushing progress.
Ian Mackintosh: I don’t feel the need to pick on particular people, but I will make a general statement. First, it has been my personal experience, that there are no more than a handful of corporations who have a really profound grasp of the benefits of creating standards and of collaboration.
Second, as Victor was saying earlier, you wish that somebody or some organization would step up and plan things. But there’s a lack of education, so it’s upon us to educate people in these things. If that’s not happening, then it’s upon us to be the missionaries. We're the problem if we're not part of the solution.
Third, I generally agree that there are groups of corporations who are culturally disinclined to cooperate. There’s a change, however, because people are starting to understand that standards are products and have to be planned out as such. When people really get it, things happen. I see Victor nodding because he and I agree about this.
Victor Berman: Yes, education is key, but you’re going up against big trends here.
Ian Mackintosh: Yes, there is all that proprietary, parochial stuff – it takes a big man to realize that you can benefit from collaboration. You can talk about cooperation, but corporations are large and diverse, so you have to figure this out in little groups within the corporation. And sometimes even the same group that’s learned the lesson of cooperation doesn’t [remember it] the next time around.
Some of us could name a few names, but there really are single individuals within the industry who actively block initiatives in this area. They’ve had their way for a long, long time here, but nature eventually will take its course. People aren’t as smart as you wish they would be, but eventually they learn from past mistakes.
Victor Berman: When Verilog was just being opened up, I came to Gateway, which was quickly swallowed up by Cadence. There was such a lot of cultural resistance there to the opening up of Verilog. We fought it out for over 18 months before the decision was finally made. Only finally did Cadence feel that they were big enough to face the challenge – but even then, opening up Verilog was like ripping the first born out of the mother’s womb. The end result, however, was that the Verilog business took off and Cadence immediately grew by 10x.
Ian Mackintosh: Yes! Standards grow markets and TAM.
Victor Berman: The market just exploded for Cadence. They went from $25 million a year to $200 million a year in their verification products. It was pretty dramatic. Although, it didn’t necessarily translate to other products.
Peggy Aycinena: Bill? Ian? Biggest gripes? Biggest hurdles?
Bill Martin: I think there are a lot of different things being discussed here. A lot of end customers want just one vendor and one solution, but that’s never going to happen. There are too many things going one for us to go off and solve everyone’s problems. Obviously, we’re in the market to make money, but you have to pick your battles.
Back at IP-06, people asked about reuse. Mentor had a market a long time ago in that space, but the market wasn’t ready. Now years later – Ralph was badgering me at IP’07 for not supporting standards, but now we’re doing a lot in that area. You have a lot of activity going on at all of the companies in this area.
We’re not solving all of the problems but we are solving some of them – particularly the key one. Synopsys is doing stuff with us, as is Cadence, as Peggy mentioned earlier. My biggest gripe is that people focus on their own gains rather than trying to increase the market. But, if we focus on growing the market, the pie will get bigger. If we focus more completely externally, rather than internally, we will do a world of good.
Peggy Aycinena: Progress is not just dependent on more golf between the CEOs? [Laughter]
Bill Martin: Yeah, standards do take some of the friction out of the market. When people know there is a network out there, and they can tie into it, the friction is reduced. It’s the same thing whether it’s GDSII, or Verilog, or tool interfaces, or how IP is created, or things like AMBA buses.
Standards remove friction. It’s like a Lego block. We don’t have that today around the world with power. There are different plugs in different parts of the world, but Sony got smart and now you can plus your Sony device into 220 V or 120 V, and you don’t have to worry about different power standards when you travel.
Peggy Aycinena: Is this a morality tale regarding the power struggle between Cadence and the Mentor/Synopsys/Magma Low Power Coalition?
Bill Martin: We have seen major changes in attitudes at both Mentor and Cadence. In the old days, as Victor mentioned, there was far less collaboration than today.
Ian Mackintosh: Certainly, and Si2 is a good example of that collaboration.
Peggy Aycinena: Will we see the emergence of organizations that provide qualitative evaluation of IP? And, is there a growing maturity in the IP market?
Victor Berman: There will be organization to do qualification of IP. There are already moves in that direction. Unfortunately, some complex IP is more difficult to deal with. There’s too much configuration, etc. But there will be more of that because of necessity, especially in the emerging companies because they’ll need it
Despite all the pain and negative examples, the industry is maturing and the value of putting standards in place is more recognized than it was in the past. There’s a certain consolidation among standards groups that’s healthy. [In fact], I see VSIA going away as a sign of that maturity.
Warren Savage: With respect to an organization that evaluates IP, no you will not see a standards body emerge to evaluate IP. Mainly because, from the standpoint of who will pay for this, the pricing pressure is on the companies to evaluate the IP themselves that they purchase.
As for reasons for optimism, we don’t have a lot. But as Victor says, the industry is maturing. There are some nice transitions from VSIA to IEEE, from SPIRIT to IEEE – these are positive signs. But we need to be cautious [with our optimism]. IP is such a dynamic and moving thing, it’s really a creative thing. It doesn’t lend itself to standard as much as other [aspects of the technology].
Ian Mackintosh: In terms of standards bodies, it’s already happening – it’s already taken place. There are experimental models where people have played around with this in Japan, China, and Hong Kong. It will expand beyond this level simply because a lot of people want to remove internal costs by sharing the costs of the work of IP evaluation. Whether it is on the table, or under the table, it’s already happening.
In terms of the maturity of the industry, I’m actually shocked at how slow we are as an industry. Ultimately, there’s money to be had from collaboration and that will drive it. That’s the only hope I see. If I just looked at the cultural and historical [trends with regards to standards], I’d be worried.
Ralph von Vignau: I don’t believe there will be a standards group that evaluates IP. There may be certification groups for those who are willing to have their IP certified. I’ll go that far, but no farther [in predicting the availability of such services]. I agree with Warren – the nature of IP is changing so rapidly!
I’m a person who’s normally optimistic. I see a major change between the standards we used to develop, and making money on those standards. Today the majority of the standards organizations are about opening markets, about making sure people can interoperate. Although, we have yet to fully prove the ROI for doing this, I believe there are standards coming along that are showcasing these opportunities for the industry.
We’re not there yet, but because there’s a heavy investment – one of the arguements about the IEEE – they even understand this now. The electronics industry is a fast-moving, changing, volatile market and you need standardization. We must do this work to survive.
Bill Martin: We work at the leading edge of IP, and I agree with Warren and Ralph. Things are too fast moving to do [formal qualification of IP], but as things get accepted you’ll be able to set up an organization [for certifying IP]. I actually had a talk with a small company about setting up a business to qualify IP – it’s a value proposition [that’s now emerging]. At some time in the future, I believe you’ll see it.
The maturity of the industry? There’s no question that the entire high-tech market continues to evolve. Back 50 years ago, the first transistor was manufactured. Now you can’t touch or use any product today that doesn’t have electronics in it. The systems are more complex, larger, and we’re talking about hundreds of millions of transistors on a chip. We can create it, design it, manufacture it, and get it into customers’ hands. The industry definitely continues to mature as a group, and IP is the next place we’re going to be going.
Peggy Aycinena: On that optimistic note, thanks for being here!
Additonal info …
Ian Mackintosh is Founder and President of the OCP-IP Association, which develops and supports standards for IP core interfaces for plug-and-play SoC design. The OCP standard is now up to Version 2.2.
Victor Berman is CEO of Improv, an IP provider specializing in DSP cores. EETimes’ Richard Goering called Victor an “EDA standards guru” last year when Victor left Cadence to take over at Improv.
Ralph von Vignau is Senior Director of the Innovation and Technology Group at NXP, and President of the SPIRIT Consortium, an organization Ralph helped found in 2006 which develops and supports specifications for meta-data and tool interfaces. SPIRIT’s IP-XACT standard “is currently under review with the IEEE Working Group P1685.”
Bill Martin is General Manager of the IP Division at Mentor Graphics, a company which is an Embedded Systems company at ESC, an IP company at IP/Grenoble, and an EDA company at DAC. Bill is comfortable in all of those venues.
Warren Savage is Founder, President & CEO of IPextreme, a company that provides licensing and market distribution support to third-party IP providers. Warren moderated the panel mentioned earlier at IP07 last December.
SPIRIT ( http://www.spiritconsortium.org)
The SPIRIT Consortium officially incorporated as an open, independent, California non-profit organization in July 2006. It announced an enhanced future roadmap covering topics such as debugging, hardware constraints, documentation, and register-description formats. Steps to create alignments with Si2 and SystemRDL in line with the future scope of The Consortium have also been announced. IP-XACT, The Consortium’s official set of specifications for IP meta-data and tool interfaces, has been released to the public and is currently in review with IEEE Working Group P1685.
IEEE P1685 ( http://www.eda-stds.org/spirit-p1685)
The purpose of this project is to provide a well-defined XML Schema for meta-data that documents the characteristics of Intellectual Property (IP) required for the automation of the configuration and integration of IP blocks; and to define an Application Programming Interface (API) to make this meta-data directly accessible to automation tools.
VSIA ( http://www.vsi.org/) … July 2007
For 11 years, the VSI Alliance was the leading IP Standards body, working with representatives from all segments of the SoC industry, easing the integration and reuse of IP. Since its inception in 1996, the Alliance released more than twenty standards, specifications and technical documents that have been widely adopted by organizations worldwide. After helping the industry sort through SoC, IP and reuse issues, the VSIA has announced plans to close and to move successful products to other industry organizations. This was a very difficult decision, particularly with the worldwide successful adoption of the QIP Metric; the tremendous industry interest in the encryption work under way; and the commercial adoption of our IP Transport specifications by Chip Estimate and Design and Reuse. The VSIA website will remain active and available to continue downloading our standards, specifications and white papers.
OCP-IP ( http://www.ocpip.org/home)
OCP-IP is dedicated to proliferating a common standard for intellectual property (IP) core interfaces, or sockets, that facilitate "plug and play" System-on-Chip (SoC) design. Making complex SoC design more efficient for the widest audience, the industry strongly supports the Open Core Protocol as the universal complete socket standard, regardless of on chip architecture or which processor cores are featured.
Accellera ( http://www.accellera.org/home)
To improve designers' productivity, the electronic design industry needs a methodology based on both worldwide standards and open interfaces. Accellera was formed in 2000 through the unification of Open Verilog International and VHDL International to focus on identifying new standards, development of standards and formats, and to foster the adoption of new methodologies.
Accellera's mission is to drive worldwide development and use of standards required by systems, semiconductor and design tools companies, which enhance a language-based design automation process. Its Board of Directors guides all the operations and activities of the organization and is comprised of representatives from ASIC manufacturers, systems companies and design tool vendors.
OSCI ( http://www.systemc.org/home)
Transaction-level modeling continues to grow in importance for architectural exploration, performance analysis, building virtual platforms for software development and functional verification. The recent release of the OSCI TLM-2 draft 2 standard marks a significant milestone in making interoperable transaction-level modeling a reality. A summary of feedback received from the public review of TLM-2 and plans for the next revision of the standard was featured at events held during the DVCon and DATE Conferences. The TLM-2 draft 2 kit is available under open source license.
Si2 ( http://www.si2.org)
Silicon Integration Initiative (Si2) is an organization of industry-leading companies in the semiconductor, electronic systems and EDA tool industries. We are focused on improving productivity and reducing cost in creating and producing integrated silicon systems. We believe that through collaborative efforts, the industry can achieve higher levels of systems-on-silicon integration while reducing the cost and complexity of integrating future design systems.
You can find the full EDACafe event calendar here.
To read more news, click here.
-- Peggy Aycinena, EDACafe.com Contributing Editor.