DAVID FINCH: This is your EETimes WeeklyBriefing. Today is Friday, April 19th, and among the top stories this week: Samsungis moving toward a 5-nanometer foundry process; Underwriters Lab iscollaborating with Edge Case Research to draft a standard for autonomoussystems. And we take a peek into Finland’s "Radio Valley" to learnabout “Life after Nokia.”
DAVID FINCH：又到了EETimes全球联播时间。今天是4月19日，星期五。本周热门新闻故事有：三星正朝着5纳米晶圆工艺迈进；UL与Edge Case Research合作起草自动系统标准。然后我们将进入芬兰的“无线谷”，了解“后诺基亚时代的生活”。
Later in the show, we'll be joined by PhilKoopman, CTO at Edge Case Research and a professor at Carnegie MellonUniversity. Junko Yoshida, EE Times' chief international correspondent, asksPhil what makes autonomy standards-- whether developed at UL or ISO-- so hardto develop.
稍后，Edge CaseResearch的CTO和卡内基梅隆大学教授Phil Koopman将加入我们的谈话。EE Times首席国际记者吉田顺子（Junko Yoshida）询问Phil自动标准包括什么？是在UL还是ISO开发？以及开发的难度等问题。
We'll hear a report by Sally Ward-Foxtonfrom our London bureau, who recently had an exclusive chat with Nigel Toon, CEOof GraphCore, the U.K.-based AI accelerator chip startup.
接下来让我们听听来自EETimes伦敦办事处的SallyWard-Foxton的报道，她最近与英国的AI加速芯片初创公司GraphCore CEO Nigel Toon进行了独家采访。
And at the bottom of the program we willtake a peek into Finland’s "Radio Valley" to learn about “Life afterNokia.”
All that to come, but first, Rick Merritt,EE Times’ Silicon Valley bureau chief, discusses Samsung’s announcement that ithas completed its work on a 5-nanometer foundry process using extremeultraviolet lithography. The announcement was a pre-emptive strike by Samsung.TSMC--the Korean giant’s largest rival-- is scheduled to give us an update onits 5-nanometer node next week. Rick explains what’s at stake in the upcomingfiner-node battle.
所有这一切都会逐一呈献，但首先，有请EE Times硅谷办事处主任Rick Merritt谈论一下三星。三星宣称已经使用超紫外光刻技术完成5纳米晶圆工艺的工作，这一消息是三星先发制人的技俩。这家韩国巨头的最大竞争对手台积电计划下周发布有关其5纳米工艺的最新消息。Rick将为我们解读即将到来的更精细工艺的竞争。
RICK MERRITT: Samsung’s foundry groupannounced this week its 6-and 5nm process nodes are ready. And this isimportant because it's the latest data points from the leading edge of work onthe traditional process of shrinking silicon that drives much of theelectronics industry.
Samsung and TSMC are in some ways in aneck-and-neck race. They're both using the latest extreme ultravioletlithography tools in production processes that offer something that's arguablycan be called 7or 6 or 5nm process. And clearly, these capabilities arebecoming more complex and costly to deliver, and that’s driving a change in howchip and system makers innovate.
So Samsung's an impressive chip maker andhas deep experience making DRAMs and NAND flash and logic. It falls short as afoundry in that it doesn't have all the kind of IP blocks and partnerships andservices that a TSMC can offer, or a size of a TSMC, which is three times asbig a foundry compared to Samsung. And it lacks the blue-chip customer listthat TSMC commands, given its status as a leading foundry.
Samsung knows where it falls short, andit's driving to be one of the leading-edge chip foundries. one of the lastcompanies standing when Moore’s Law finally crashes into the atomic limits,perhaps around a 2nm node, let's say in 2024.
This is Rick Merritt in Silicon Valley forEE Times On Air.
DAVID FINCH: UL’s move to take a plungeinto autonomy standards might have surprised some in the engineering community,especially those of us more seasoned.
Junko sat down with Phil Koopman, CTO ofEdge Case Research and author of the UL 4600 draft. Phil initially approachedEE Times to reveal what has been going behind closed doors at UL.
JUNKO YOSHIDA: Hi, Phil. Thank you forcoming to the show.
So we learned from you this week thatUnderwriters Lab, UL, in close collaboration with Edge Case Research, isworking on UL 4600 for autonomous systems. In parallel, we reported last monththat ISO is working on a separate standard called SOTIF for ADAS and Autonomousvehicles.
Here’s my first question, Phil: What makesany autonomy standards-- whether done at UL or ISO-- difficult to develop?
PHIL KOOPMAN: Well, Junko, any autonomystandard has to grapple with the fact that this is a rapidly evolvingtechnology area. The last thing you want to do is have a standard for eachprogress. ISO 21448 SOTIF is at its best when you have a set of requirementsand you can apply more or less traditional safety engineering approaches. Thatmeans they concentrate on squeezing out as many of the unknowns as possible.And that way, what you're left with are unknowns that you can apply traditionalsafety to.
But as you move out of ADAS into fullautonomy, the UL 4600 draft goes further in a couple directions. First, itencourages managing the risk of unknown unknowns based on continual fieldfeedback. So you may ship with an unknown, but you're going to find out aboutit and fix it promptly to make sure the risk is minimized.
Second, it helps make sure the designershave thought of all the not-to-obvious implications of removing the humandriver. A lot of 4600 is just a list of things to remember to think about, sodesign teams don't have to relearn lessons the hard way that other industriesand other companies have learned. In other words, it helps you avoid puttingvehicles at unnecessary risk.
JUNKO YOSHIDA: Now, here’s what I thinkgets interesting. Once it is not a human being but a machine that is driving anautonomous system, there are a lot of things system designers must think aboutin advance and prepare for safety.
In our interview, when you and I talked,you mentioned a couple of times, “Did you think of that?” Give us those “did you think of that?”examples please.
PHIL KOOPMAN: We found a lot of these, butI can give you a couple of examples to give you an idea.
First one: In a regular car, designers cantake credit for the driver doing the right thing if something goes wrong. Forexample, if you lose your normal braking system, the driver is supposed to usethe parking brake to stop. And by the way, this actually has happened to me,and I did use the parking brake to stop. If you put an autonomy system on a carplatform like that, I certainly hope that the autonomy system knows how to dothe same thing. What this means is that if you're combining an ISO 26262approach with autonomy, all the credit that got taken for the human driverdoing the right thing-- in other words, 26262 controllability-- now gets heapedon to the autonomy.
A second, perhaps more subtle example isthat things in the supply chain, or even small maintenance mistakes, can gofrom being an annoyance to life-critical. As an example, let's say that someonein the supply chain waters down the windshield wiper fluid. They figure they'regoing to sell them summertime and it's no big deal if it doesn't have enoughantifreeze in it. So you refill your car in the summer, and things are fine fora while. Or maybe you don't go down to the store and get fluid, you just putsome tap water in. Either way, ends up the same place. But along comes winter,and some mud splashes on the windshield and the wiper fluid is frozen so youcan't clean your windshield. Not a lot of fun in a human-driven car, butsomething like this has probably happened to a lot of us, and we figure it out.But if you're in an autonomous vehicle, and let's say you're the passengerasleep in the back seat, and all of a sudden all the cameras and the lidars arecoated with mud, and that wiper fluid isn't there to clean things off. Nowyou've got a problem. You've got a vehicle going at speed that can't see.
Now I'm not saying that you have to solveit one way or another. Maybe there's some radar; maybe you have some otherplan. That's great. But the question is, "Did you think of that?" Areyou sure that you're going to be safe when that happens? Because it willeventually happen.
JUNKO YOSHIDA: UL 4600-- the draft you areworking on-- is said to cover “validation of any machine learning-basedfunctionality, and other autonomy functions used in life-criticalapplications.” Now here's my third question: Can you actually validate machinelearning-based autonomy, and if so, how?
PHIL KOOPMAN: We've got an industry bettingtens of billions of dollars that they can pull it off. So I certainly hope so!But to answer your question, Junko, the good news about the UL 4600 approach isthat it gives companies a lot of flexibility. So long as the story they tellmakes technical sense and they can say, here's why we're safe and here's whyyou should believe us, that's great. But this is a two-edged sword. To get allthat flexibility, that means they have to decide how they want to argue they'resafe.
So I expect that's going to entail cleverlylimiting the operational design domain, and probably it will involve having asystem that's pretty smart about knowing when it doesn't really know what'sgoing on, and so it can be cautious.
It also talks about good safety hygiene andthe types of things that are going to make it easier to explain why you'resafe.
But beyond that, UL 4600 does not tell youhow to build the system, it tells you that you have to explain why you're safe.Actually getting there? That's up to the designers.
JUNKO YOSHIDA: Phil, thank you so much for coming to theshow. We really appreciate it.
DAVE FINCH: And now to our London bureauwith Sally Ward-Foxton, who shares what she learned from GraphCore’s CEO. Theoutspoken executive now claims, “GPUs are effectively holding back innovationin developing futuristic neural networks in AI.” Here's Sally with more.
SALLY WARD-FOXTON: GraphCore are a Britishstartup working on AI accelerator chips. They are an intriguing company, partlybecause they’re said to be the only semiconductor unicorn in the western world.At the last funding round, their company valuation was 1.7 billion dollars.Their investors include Dell, Bosch, BMW, Microsoft and Samsung, amongstothers, and they’ve raised 300 million dollars so far. So obviously investorssee something exciting in what GraphCore is doing.
The other reason GraphCore is so intriguingis their technology. They’ve been working with some of the leading minds inneural network research, and they’ve built a processing architecture speciallyfor AI. It’s called Colossus, on account of it allows for building absolutelyenormous processors. One GraphCore chip, one intelligence processing unit withthe Colossus architecture, has 1200 separate cores, with each one running sixprogram threads, so each chip can deliver 125 teraFLOPS of computing power. Thenumbers are absolutely staggering.
There are several important things I cameaway with after talking to GraphCore’s CEO, Nigel Toon.
First was the realization that neuralnetworks are very diverse. Just as machine learning is applied to very variedtypes of tasks, neural networks for those tasks are obviously very different.When we’re talking about building a general purpose chip that can accelerateall different types of neural network, it’s much harder than building chips toaccelerate specific AI applications. It’s this general purpose accelerator chipthat GraphCore is working on.
Secondly, GraphCore is looking not just atthe present, but towards the future of AI. As deep learning evolves, we’ll usetechniques like reinforcement learning, where data is fed back through thenetwork more than once, so the system can learn from experience. This isimportant when we want it to try and learn something about context, maybe if itwas understanding spoken conversations, for example. As a result, we might endup with less separation between training chips and inference chips, because theinference side of things will be expected to learn as it gains more experience.
Finally was the idea that the incumbenttechnology for machine learning, GPUs, are effectively holding back innovationin developing these futuristic neural networks. Nigel Toon made the point thatGPUs are great if you’re doing a relatively basic feed-forward convolutionalneural networks, maybe looking for certain objects in photos. But for emergingtechniques like recurrent neural networks and reinforcement learning, theydon’t map well to GPUs. He said that whole areas of research are effectivelybeing held back because there isn’t a good enough hardware platform available,and the fact that people are building application-specific chips and playingwith FPGAs for machine learning research is a clear sign that GPUs aren’t up tothe task. Hence the need for GraphCore’s IPU processor.
GraphCore, of course, like any startupcompany, they face several challenges. They’ve built a massive war chest, butthey need to be focused on getting return on investment for their investors.Part of the problem will be growing quickly enough. They are currently at 200people, and Nigel Toon says they’ll look to double in size this year. They havethe cash, but finding that amount of qualified silicon engineers and software engineersin the space of a year will not be an easy task.
This is Sally Ward-Foxton reporting fromLondon for EETimes.
DAVE FINCH: And finally, our Londoncorrespondent Nitin Dahad recently visited Oulu, Finland.
We often wonder what happens to a companytown, anywhere in the world, after the anchor company closes its R&Dfacilities, shuts down its factories and lays off thousands of workers.
Here's Nitin's report on “Life after Nokia”in Oulu.
NITIN DAHAD: As a visitor to the city ofOulu in northern Finland, with its 200,000 population, three things defined itfor me (apart from the heavy snow): the university, Nokia and saunas.
Yes, Finland is the land of saunas, with 2million saunas for 5 million people, and it's an integral part of daily life, whichI’ll come to later.
Oulu is a small city where a lot ofbusiness activity happens within something like a 20-mile radius.
So, when Nokia closed its mobile phonebusiness around 2012, it was almost like an earthquake hit the city, as some7,000 people lost their jobs. It was a real shock to the system, since the citywas highly dependent on Nokia.
And even worse to come was the closure acouple of years later of Broadcom’s and Microsoft’s facilities.
I mentioned the University of Oulu. Itsstrength in radio technologies and microelectronics have been a key foundationfor innovation, both in the Nokia era and the post-Nokia era.
I saw that with the work they are doing inboth 5G and beyond (which people are now starting to call 6G). And it’s not justoperating alone. It has a collaborative network of public and private partners.
This in my view led the foundation for thecity’s regeneration.
I’ve been involved on and off in economicdevelopment for many years, and in the late 1990s, governments around the worldstarted looking at the success of Silicon Valley in California and trying tofigure out how they could emulate the same formula and create the next SiliconValley in their own countries.
It’s common wisdom that the key ingredientsfor the melting pot are knowledge, talent and money.
In Oulu, the university provided theknowledge foundation. And the talentcame from the experience gained at Nokia over many years.
The next bit was the money, and that’swhere there appeared to be (at least this is what city officials tell us), acoordinated program from Nokia and local government to fund startups who hadideas.
The whole region effectively became a radioincubator, with entire teams who were laid off starting their own companies--such as Testilabs, who bought out lab equipment from Nokia and started offeringradio test services to the industry as an independent service. Others includeactive noise canceling earplugs developer Quieton, and Haltian, a hardwaredesign engineering firm focused on IoT.
You hear many such stories as you wanderthrough the corridors of some of the office and lab units in Oulu, which wereonce themselves home to Nokia’s engineers.
I mentioned three ingredients forsuccessful economic development: knowledge, talent and money. But if it was soeasy, every city in the world would be able to do it. There’s more to it: youneed the networks and the culture, as well as customers.
The ex Nokia-staffers had strong networksand a culture that enabled them to collaborate when outside Nokia. This isprobably where the saunas come in. Saunas are an integral part of the way ofdaily life of Finnish people. KNL Networks, for example, was established afterthe founders had the idea while sitting in a sauna.
Saunas aside, I think it creates strongbonds of mutual trust among people, and this is what ultimately is essentialfor doing business.
So when outside Nokia, people were able tocall upon their networks to do the design, the validation and testing, and themanufacturing. And be sure that services and products would be deliveredaccording to market needs and when they needed it.
Many of the startups coming out of Nokiawere able to reap the rewards of these networks quickly, some getting contractswith key blue-chip customers around the world within a year. This created arich ecosystem of companies in the entire supply chain in Oulu, creating what Ilike to coin a "Radio Valley."
The regeneration and development model Imentioned is repeatable, and it's been done with varying degrees of success.Israel’s Tel Aviv and Haifa are probably the most well-established andsuccessful examples. Then you have Bangalore, London, and in my travels, I’veseen similar emerging clusters in places like Cairo and Alexandria in Egypt,Sao Paulo in Brazil, and many others around the world.
This is Nitin Dahad, EE Times’ EuropeanCorrespondent, reporting from Oulu, Finland.
DAVE FINCH: That was Nitin Dahahd, and thishas been your weekly briefing from EETimes and the AspenCore Global Service.You can read all these stories and more at EETimes.com. Thanks for listening.