波音737 MAX空难国会听证、Semicon West芯片增长预测、Workstation是什么?

BRIAN SANTO: I'm Brian Santo. I'm the Editor in Chief of EETimes. You're listening to EETimes On Air. This is your weekly briefing for the week ending July 12th.

BRIAN SANTO: 我是EETimes主编Brian Santo。你正在收听的是EETimes联播, 以下是截至712日的一周新闻热点播报。

Our stories this week:


You might recall pilot Chesley Sullenberger. He became a hero a few years back for crash landing a disabled passenger airliner into New York's Hudson River with no loss of life. He recently appeared in front of Congress to testify about the crashes of the Boeing 737. We discuss his testimony and what it means for Boeing and for other engineering companies.

你可能还记得飞行员Chesley Sullenberger。几年前,他因驾驶一架出现故障的客机成功降落在纽约哈德逊河上且没有造成任何死亡而成为一名英雄。最近,他在国会听证会上就波音737的坠机事件作证。我们讨论了他的证词及其对波音和其它工程公司的影响。

We've got an on-site report from Semicon West, including a revised estimate of growth in the chip market in 2019.

接下来我们收听来自Semicon West的现场报道,包括2019年芯片市场增长的修正预测。

And guest commentators John Petty and Kathleen Ma have just concluded a major report on the workstation market, which has seen some profound changes over the years.

客座评论员John PettyKathleen Ma刚刚完成一个有关工作站市场的重要报告,由此可以看到这些年来发生的一些巨大变化。

KATHLEEN MA: Now it is a workstation.

KATHLEEN MA: 现在它是一个工作站。

JOHN PETTY: Oh, I don't know. It's almost whatever you want to call it.

JOHN PETTY:哦,我不知道。你想称呼它什么都可以。

BRIAN SANTO: We'll get back to John and Kathleen in a moment. First up the Boeing 737 disaster.

BRIAN SANTO:我们稍后再跟JohnKathleen讨论,来看看波音737难。

Earlier this year, there were two separate crashes involving the craft. There is an ongoing investigation into precisely how those crashes happened, and perhaps more importantly, why they happened. Indications are that Boeing was rushing to redesign the plane in such a manner that it could avoid new inspections by the Federal Aviation Administration. Congress recently held hearings to look further into the matter. The testimony by Chesley Sullenberger during those hearings was particularly dramatic. Editor George Leopold has been following the story for EETimes.

今年早些时候,波音737机型发生了两起坠机事故。目前,一项针对这些坠机事件真正原因的调查正在进行中,更重要的是要查出为什么会发生坠机。有迹象表明,波音公司急于重新设计这款机型,以避免联邦航空管理局(FAA)展开新的调查。国会最近召开听证会以进一步调查此事。Chesley Sullenberger在听证会上所做的证词特别引人注目。EETimes编辑George Leopold一直关注这一事件的进展。

So George, tell me what the significance of Sullenberger testifying.


GEORGE LEOPOLD: Well the House Transportation Aviation Subcommittee held this hearing on I think was on June 19th, and the idea was to bring in the stakeholders – meaning the people who have to fly the aircraft, I think even flight attendants, and I think they even had some passengers. But obviously Chesley "Sully" Sullenberger is, you know, he's beloved in this country. He's the guy who survived an unprecedented loss of two engines and he put down his airplane on the Hudson River and saved 155 people, 154 counting himself. They all walked away. So he's got some street cred, or maybe we should say sky cred.

GEORGE LEOPOLD: 好的。我记得众议院交通航空分委会在619日举行了一次听证会,意在邀请相关方,也就是那些驾驶这一机型的飞行员,我想还有空乘人员,甚至一些乘客。显然,ChesleySullySullenberger在美国深受人们喜爱。在两台发动机毫无征兆地失灵的情况下,他将飞机安全迫降在哈德逊河上,使得155人得以生还,包括他本人在内。所有人都幸免于难,因此他获得了街头巷尾的赞誉,或者我们应该称之为空中赞誉


But his testimony was damning, and he revealed during his testimony to the subcommittee that he has flown the 737 MAX simulator. He flew the two flight profiles of the Lion Air and Ethiopian Airlines crashes, and he testified that he could see by the conflicting warnings he was getting from the angle of attack sensor, this what looks like a single point of failure, that the pilots could have been very confused, didn't have enough time and didn't have enough altitude to pull this thing out of this scenario.

但他的证词颇具谴责意味,在向分委会作证时透露,他已经使用了737 MAX模拟器飞行。他模拟驾驶了印尼狮航和埃塞俄比亚航空公司两起空难的飞行模式,并作证说,他从瞄准角传感器获得的警告相互矛盾,这似乎像是故障的关键点,飞行员可能会很困惑,没有足够的时间,也没有足够的高度来将飞机拉出这个(危险)场景。

BRIAN SANTO: So this is something that's new to me. I was unaware that it's possible to record what happened in a flight and then simulate the actual flight. The fact that Sullenberger was actually able to put himself in a simulation of what actually happened with the two downed jets.

BRIAN SANTO: 这对我来说倒是新消息。我还不知道根据飞行记录中实际发生的情况来模拟真实的飞行。Sullenberger实际上能够通过模拟飞行将自己置身于两架坠毁飞机之中来看到真实发生的情况。

GEORGE LEOPOLD: Well I think, you know, if they find the black box recorders, they'd probably have a sufficient amount of telemetry that they can take it to a simulator and get pretty close to what the pilots saw and felt. And what he says they saw and felt was one warning called a stick shaker saying, "You're going too slow; you're going to stall." And simultaneously, another contradictory warning, a clacker, is saying, "Too fast! Too fast!" If anybody saw the movie "Sully," you know, you got a sense of the warnings that are going on, and you got to figure out really fast what's going on with your aircraft and what are you going to do to get it back to the runway.

GEORGE LEOPOLD: 要知道,如果找到了黑匣子记录仪,他们可能会有足够的遥测信息,可以将它导入模拟器,非常接近地还原飞行员所看到和感受到的情况。Sully说道,飞行员们所看到和感受到的是一个称作摇杆的警报:“你的速度太慢;你要熄火了。”与此同时,却有另一个相反的噼啪声在发出警报:“太快了!太快了!” 如果有人看过电影“Sully (萨利机长)”,就会明白正在发生的警告,你必须非常快速地弄清楚这架飞机正在发生什么,你要怎么做才能让它回到正常航行状态。

BRIAN SANTO: Well I've got to imagine that even when you're trained, that's got to be terrifying when you're getting conflicting alarms.

BRIAN SANTO: 好吧,我可以想象,即使你受过训练,当你收到相互矛盾的警报时,也会感到非常可怕的。

GEORGE LEOPOLD: Yeah. He even noted he even knew what was coming and he still couldn't pull out of this death dive that both these aircraft were in.

GEORGE LEOPOLD: 是啊。他甚至指出,即使他知道将要发生什么,他仍然无法摆脱这两架飞机的死亡下冲。

BRIAN SANTO: Wow. Wow. That's crazy. So between Sullenberger's testimony and, you know, the time lapse in between then and now, it's been a matter of 10 days or so, has the story moved forward any? Have we learned anything new about the MCAS system or what's going on with the investigation into Boeing?

BRIAN SANTO: Wow,这太疯狂了! 那么自从Sullenberger作证到现在,已经过了10天左右的时间,事件是否有新进展? 我们有没有关于MCAS系统的新消息,或是针对波音公司调查的新情况?

GEORGE LEOPOLD: Yeah. The latest is, they've come up in simulations with what looks like a hardware problem, a processor problem. And as I understand it, what they were trying to do is basically have the software activate the switches in the flight control system, and they wanted to introduce a processor into this system to make the computer faster. And according to the reports I've seen, this hasn't been working too well, and it looks now like it's another problem with MCAS.

GEORGE LEOPOLD: 有的,最新消息显示,在模拟飞行中出现的情况看起来像是硬件问题,一个处理器问题。据我了解,波音试图做的事情,基本上讲是让软件激活飞行控制系统中的开关,并且他们想为该系统增加一个处理器,以便计算速度更快。据我所看到的报道,这个并非如期望的那么好,现在看起来它是MCAS的另一个问题。

The other thing that I've heard – I've not been able to confirm – is that Boeing may have outsourced the development of the MCAS system overseas, which seems a strange thing to do. Again, Sully Sullenberger said the people running these programs ought to be pilots so they understand, you know, what these systems have to do when you're pitching stabilizers up and down and, you know, what that does for an airplane that's climbing to the cruising altitude.

我还听到另外一件事,但还无法证实,即波音公司可能将其MCAS系统的开发外包给国外公司了,这似乎有点奇怪。Sully Sullenberger重申,开发这些程序的人应当是飞行员,这样他们才能理解,当向上/向下推动稳定器时,系统必须做什么,以及往飞行高度爬升对飞机意味着什么。

And then the other thing I'd point out is that the Seattle Times has done great coverage on this, and they had an inside account of how this all came about, how MCAS came to be. And I link to it in our story and recommended it to our readers. And people, the sources we talked to, have also recommended it highly.


BRIAN SANTO: Fantastic. So this is Boeing's very specific event, but the problem of a lack of controls in the manufacturing process, lack of foresight in the manufacturing process, perhaps the pressure on a company to get something out before it's fully baked, that's not uncommon. Are there any consequences for electronics companies or other industrial companies at large coming out of this?

BRIAN SANTO: 很好!虽然这次是波音公司的特定事件,但这种情况并不少见,包括制造过程中缺乏管控和预判的问题,也许这是一家公司在产品没有完全成熟之前就推向市场所面对的压力。这次事件对电子公司或其他工业公司会带来什么影响?

GEORGE LEOPOLD:  Well I think – and if it happens it'll be a good thing – I think there’ll be closer scrutiny of how the components are manufactured, how they're installed. I mean something else has cropped up in the last week or so where Boeing's got problems at their 787 Dreamliner factory in South Carolina where they found all kinds of shoddy manufacturing. And it's very simple things like tie-downs so that bolts won't pop out. And maybe Boeing, as an attempt to save itself, moves some of its manufacturing back to Renton, Washington, so they can keep a closer eye on these workflows. So that's one possible outcome.

GEORGE LEOPOLD: 我想如果会产生一些影响,那将是一件好事。我认为这将使得元器件的制造和安装方式的审查更加严格。我的意思是,在过去一周左右,出现了一些其他问题,波音公司位于南卡罗来纳州的787 Dreamliner飞机工厂遇到了问题,他们在那里发现了各种伪劣的制品。这是非常浅显的道理,拧紧螺栓以免螺栓松脱。也许波音为了自救,将其部分制造转移到了华盛顿州的伦顿市,以便他们能够密切关注这些工作流程。这是一个可能的结果。

And you know I guess the other thing is, will this aircraft ever be recertified to fly? And that's at this point it's not clear it will. And I'm not sure a lot of passengers will want to fly on it.


BRIAN SANTO: Speaking as a potential passenger? Yeah, that's a consideration, isn't it?

BRIAN SANTO: 你是作为潜在乘客在发表观点吗?对,这也是一个考虑因素,不是吗?

GEORGE LEOPOLD: Yeah. I know. Yeah. I know their website, they're going to play up the fact that your equipment is a 737 MAX. Do you still want to make this reservation? So that down the line that's going to hit them.

GEORGE LEOPOLD: 是啊。如果你在机票预订网站看到你的执飞机型是737 MAX,你还想预订吗?这对波音来说打击真的很大。

BRIAN SANTO: Wow. Crazy. Well George, thanks for being here with us this week.

BRIAN SANTO: Wow,真是疯狂! 好的George,感谢本周和我们一起讨论。

GEORGE LEOPOLD: Good to be back with you, Brian. Take care.

GEORGE LEOPOLD: Brian,很高兴和你交谈,再见。


BRIAN SANTO: Semicon West is the annual conference held by semiconductor factory equipment suppliers. EETimes editors Rick Merritt and Dylan McGrath were at the show, and they filed this report from the conference hall.

DYLAN McGRATH: Hi, this is Dylan McGrath along with Rick Merritt. We're here at the Semicon West in San Francisco this year. Rick, how are you?

RICK MERRITT: I'm doing great. Now you've been here for 20 years. What's changed?

DYLAN McGRATH: Well, I was just thinking, the show has changed quite a bit. It's gone from big exhibition to a much smaller exhibition, and I spent much of my time in the AI design forums. You know, there's just a lot more content here than there used to be. A lot more presentations and less focus on what's going on on the show floor. But you did walk the show floor. What do you see out there?

RICK MERRITT: Well, yeah, it's really busy. There's all those small and medium companies from the capital equipment sector as you well know. The big, the small and medium ones. There are valves and tubes and everything. But as you know, it is better than anybody. How is that sector really doing right now? I mean it looks like a party on the show floor, but how are they really doing?

DYLAN McGRATH: Well it's not a party as you know. So I spent a lot of time in presentations yesterday with analysts, and yeah, things are not looking good. I think everyone understands that. What I took away from yesterday was, things are actually looking worse than people thought they were, especially at the start of the year. There was some expectation that the market for chips might be down a little bit. Now it's looking like 10 percent plus it's going to contract. And the latest estimate is that the equipment market is going to contract by 18 percent.


DYLAN McGRATH: While the drop in memory chip prices, everyone saw that coming. But there's going to... it looks like this is going to be a very significant downturn.

RICK MERRITT: Yeah. It's the memory drop, it's the data center pull back because they overspent the last two years, and it's also the the whole trade war going on.

DYLAN McGRATH: That doesn't help. Although people are still kind of reluctant to talk a lot about the trade war, it's kind of the elephant in the room. Nobody knows what's going to happen with that, right?

RICK MERRITT: And maybe the big technology issue is, you know, we debated whether or not Moore's Law is ending. It doesn't matter. But the point is, CMOS scaling is slowing and getting expensive. So we're back to... maybe we're not hitting a wall, but we're really coming to a slowdown in a big way. So what's the big hope for this?


Yeah. Well that's one of the big takeaways that I got from that this morning at the AI design forum is, whether or not Moore's Law is truly dead. It is no longer going to drive the industry as it once has. And Gary Dickerson, the CEO of Applied Materials, said, We can't continue to do things as we've done them before. So there are heterogeneous architectures, there's new materials, new device structures. I mean you name it. They're looking at everything or looking at how they can collaborate to build things a lot faster than they used to.

RICK MERRITT: Yeah. And the good news is, people from Google get up and say, Look we've just built three generations of AI accelerators. We've got whole buildings with our own chips. We need the next generation faster. Guys, you need to develop some new transistors or something. So here's the guys with the big pockets that look like the pole vault so these guys get over Moore's Law ending and get on to wherever the next thing might be.

DYLAN McGRATH: Yeah. And they've got the money. As you say, the people do what they want. So that's that's the good news.

RICK MERRITT: So there's a short take from Semicon West.

BRIAN SANTO: John Petty and Kathleen Ma are principles at John Petty Research. They recently concluded a massive study on the workstation market, examining how workstations have evolved over the years and how they're being used today.

KATHLEEN MA: Hi, John. First of all, let's talk a little bit about the history of workstations, because they had a very clear-cut beginning, but it's gotten a little muddier as time's gone along.

JOHN PETTY: Well, it was clear-cut in that the workstations were so dramatically differentiated. They had risk processors, they had custom graphics, they had Unix operating system. They were big. They were expensive. And they didn't sell a lot of them.

KATHLEEN MA: No. And then came along the x86 and then professional versions of operating systems from Windows. And we got the famous Win telexes. It did accomplish moving, or killing, the traditional workstation market and bringing along this whole new class of computers, something between PCs and workstations.

JOHN PETTY: Yeah, because they figured out how to apply commercial and consumer processors, which were being made in huge volumes and so they were less expensive. And once they crossed that threshold, then it was, you know, forget about the big guys the way they were. And the transition was really tough for a lot of companies. A lot of companies just don't exist anymore because they couldn't make the transition.

KATHLEEN MA: Well you know, what's amazing in the work we've been doing recently, we're seeing that that transition is really still going on.

JOHN PETTY: Yeah, yeah, it is. That's right.

KATHLEEN MA: So NOW what is a workstation?

JOHN PETTY: Oh, huh. I don't know. It's almost whatever you want to call it. We have multiple processor choices now. We have the traditional x86, and that comes in two flavors, which is core-i something 3 5 7 9, or it comes in a Xeon. And then AMD has their Epic. So there's, you know process for choices like crazy. We're concerned about memory, size how much of the model we can get stuck in the large memory. There's a question about, do you need error-correcting memory or not? Yes or no? It's a mixed answer. There's no solid answer on that. Security? Yes or no again. Rugged construction? Sometimes yes, sometimes no. I mean, so it's... there's a very big gray line, if you will, between workstation users who need everything that a traditional workstation would have versus those who can take an off-the-shelf PC.

KATHLEEN MA: What I found is, I did some research just poking around the relationship between the users and the applications. So what do CAD users gravitate towards and what do game developers gravitate towards?



KATHLEEN MA: And there's big differences there. But it was surprising to realize it's still most professional applications aren't threaded for multi-threaded processors, so it tends to be you don't want the super-duper workstation, multicore workstation processor. You what to gravitate towards something faster. And all those kinds of considerations. And a lot of users are skeptical of the need for workstation graphics board versus a great gamer board.

JOHN PETTY: And also there's the discussion about error-correcting memory. That's one of the distinguishing features of a workstation-- by our definition of a workstation-- is that it has error-correcting memory. Well you can only get error-correcting memory it uses a Xeon processor. You can't get that with the consumer processor...

KATHLEEN MA: Or an Epic processor.

JOHN PETTY: Or an Epic. Yeah. So there's that. And then the other question that keeps coming up is, do you need application certification or not? And so I contend that, you know, serious workstation users do. But there's a whole bunch of people who don't care.

KATHLEEN MA: In some cases of users, I think I've seen it a lot in the CAD market, is that they don't hold their vendors necessarily in such high regard. So they're not so sure, you know, what good certification is. And they've been doing it for years. It's like, "I've gotten by so far."

JOHN PETTY: Yeah. One of the things that we've found is that people are buying machines that are labeled "workstation," but don't have any of the hardware that should be in a workstation.

KATHLEN MA: You did that survey recently that I thought was so interesting. We asked people what do they think should be in a workstation?

JOHN PETTY: Yeah. What did we find out?

KATHLEEN MA: They weren't so sure.

JOHN PETTY: What we found out was that if you don't know what's in the workstation. you don't need one.

KATHLEEN MA: I think that sums it up pretty well, John.

JOHN PETTY: Yeah. Yup.


JOHN PETTY: Everybody go out and get a workstation.

BRIAN SANTO: Now for this week's blasts from the past.

On July 8th 1957, Control Data Corporation was founded. For years, CDC built the fastest supercomputers in the world.

On July 9 1982, Disney released "Tron." It was the first major motion picture to take place in what we would now call virtual reality.

On July 11th 1976, the company K&E made the last mass produced slide rule, which it presented to the Smithsonian Institution.

On July 10th in 1962, Britain, France and the United States launched the Telstar communications satellite into space. The orbiting of the satellite inspired a British group to record an instrumental number in its honor. It was the second British single to reach number one in the US, and it seems to have been the first engineering-inspired song to top the charts ever. Here's The Tornadoes with "Telstar."


BRIAN SANTO: And that's your weekly briefing for the week ending July 12th.

This podcast is produced by AspenCore Studio. It was engineered by Taylor Marvin and Greg McRae at Coupe Studios. The segment producer was Kaitie Huss.

The transcript of this podcast can be found at eetimes.com, complete with links to the articles we referred to.

We'll be back next Friday with the next episode of the weekly briefing of EETimes On Air.

I'm Brian Santo.

感谢收听本期推送,全球联播 (EE|Times On Air) 现已同期在喜马拉雅以及蜻蜓FM上线,欢迎订阅收听!