A No bullsh*t/down to the nitty gritty perspective on embedded chips

greenspun.com : LUSENET : TimeBomb 2000 (Y2000) : One Thread

A good read and about time.


-- Goldi (goldilucks@yahoo.com), October 22, 1998


Hmm..this article is anything but professional. The guy rambles on and on and doesn't say anything but "there is no embedded chip problem, it's bull****", and doesn't explain his position within the first half of the article, which is when I stoped reading.

Anyway, here's a conterbalance to your article, The Practicle Engeneer, Making Embedded Systems Year 2000 Compliant by Dick Lefkon. He uses a specific system, security systems, to explain the problem. There's several graphics interspersed , for example one showing exactly on a chip which pin is at fault. Although he uses some techy jargon, the overall article is very well written and understandable by us lay people.

-- Chris (catsy@pond.com), October 22, 1998.

oops...cut/pasted the wrong link, here's the right one. Sorry.

-- Chris (catsy@pond.com), October 22, 1998.

Harlan Smith -- who is very well respected on embedded systems -- is replying to another individual who asked him to review an initial article. Chris missed a good article, and then criticized it because the engineer who wrote it wasn't a professional writer.

What a lot of people simply have refused to grasp -- and what Harlan is trying to emphesize -- is that THERE IS NO EMBEDDED CHIP PROBLEM, but there is a very severe embedded system problem. This is what Harlan points out.

What's the difference? A lot. Most 'embedded systems' are made up of several chips. Your PC is a small system. Power it down, pull the case off of it, and count the chips on the motherboard at the bottom. Now, one of them is the main processor. It requires the rest of the chips in order to function as a microprocessor.

Chips aren't the problem --- until they're connected together to form a system. Beware of the.......system.

Is this semantics? No. A chip is a chip, and embedded system is a system composed of several chips. One of the real problems that is going to occur is that -- if a system is non-compliant -- the entire system may have to be replaced because a form, fit, function chip replacement can't be found. This may result in severe supply problems. Roleigh Martin


had indicated that a supply problem may occur when everyone in the power industry, for example, wants to replace equipment at the same time. The more pieces you have to replace at one time, the greater the potential that one or more items won't be available.

It is a good article.


-- rocky (rknolls@hotmail.com), October 22, 1998.

Good show Rocky

You sir, "shine out like a shaft of gold, when all around is dark." Seriously though, I thank you for the insight, makes a lotta sense.

-- Uncle Deedah (oncebitten@twiceshy.com), October 22, 1998.

Thank you for clarifying his position. I guess I lost patience too soon. You have to admit that it's a lot of rambling and repeating of himself. (Reading such articles when it's way past my bedtime will do that to me.)

I already had grasped the difference between the "embedded chips" and "embedded systems" count difference, very early on in my own awareness phase, mainly from articles like this one http://www.tmn.com/~frautsch/y2k2.html (2 others I had bookmarked are not there anymore). Although this perticular article mentions the Gartner Group's chip count, it also very clearly emphasize the "embedded systems", or "black boxes", systems within systems, each potentially harboring a non-compliant chip that could cause the entire system to malfunction, as one other article explained.

Blah...my brain's too fuzzy to discuss this, I'm over my head and should have been in bed ;)

-- Chris (catsy@pond.com), October 23, 1998.


But surely he is more valuable than gold, for gold can do do nothing but reflect the light falling it. Without light, gold is but a cold hard metal....

He _is_ the ray of light, illuminating the darkness of our ignorance in the midst of our forthcoming trooubles. For it is only by illumination that we (lost in the darkness) can find our golden treasures amidst the trash .....

-- Robert A. Cook, P.E. (cook.r@csaatl.com), October 23, 1998.


In reference to your reference about my reference, check with Richard Dale. Or any Python psychopath.

-- Uncle Deedah (oncebitten@twiceshy.com), October 23, 1998.

PS. It involves bats. (the flying kind)

-- Uncle Deedah (oncebitten@twiceshy.com), October 23, 1998.

I appreciated Harlan Smith's critique and as he points out, it is still "A damn big problem".

Some companies have taken this seriously and some have not. We are still pretty much in the dark. Some services will fail, some will not. Exactly which truly mission critical services will fail and for how long they will fail is not any clearer.

But it is precisely this information that we need in order to make informed, educated decisions. We need it to make measured contingency plans. We need it to avoid panic. We need it to avoid doing something stupid. We need it to adequetly protect our families.


-- Arnie Rimmer (Arnie_Rimmer@usa.net), October 23, 1998.

Isn't it because of the "black box" type application of embedded systems that this issue becomes such a tremendous problem? Each application, though called exactly the same, may include different chip sets etc from box to box, even if it comes from the same manufacturer. Is that correct?

Also, is it because many of the vendors no longer exist and also ,because of the specialized nature of the systems, they may not be easily replaced.

Is that the jist of it? _______________________________________________________________

-- Do I have a clue? (give me a@clue.com), October 23, 1998.

The referenced article seems to contradict some of the things about embedded chips that I thought were pretty much established -- like yes, you can have a single chip, indeed it can have a date/time function that is running internally, and yes it can fail or start doing undesirable things if it cannot represent the date/time properly. See http://www.tmn.com/~frautsch/y2k2.html for a full explanation (replete with a lot of citations to other documentation -- something woefully lacking in the referenced article).

-- Jack (jsprat@eld.net), October 23, 1998.

In fact, after re-reading the article that I just posted the URL to, I sure would say that there IS an embedded chip problem, at least as far as us lay people are concerned. It may be that it is legally accurate -- uh, excuse me, I mean technically accurate -- to say that only when chips are arranged to form a system do they represent the potential for being a problem, and therefore its an embedded SYSTEMS problem. But, c'mon, thats hair splitting.

-- Jack (jsprat@eld.net), October 23, 1998.

I understand and agree with the distinction being made because I come at this from the technical side. To the people tasked with identifying and correcting potential Y2K problems, the distinction is an important one which may or may not be helpful depending upon the specific situation. For example, replacing the entire 'system' might offer a better approach than identifying and correcting a smaller embedded 'subsystem'.

But to the average layperson, the distinction between hardware, software, board-level systems, chips, PALs, ICs, VLSICs, interrupts, buses, firmware, controllers, microcode, drivers, etc. that comprise such 'systems', is essentially meaningless. The important thing to remember is that these systems are frequently embedded within other systems which are themselves part of larger 'systems' and so forth.

These 'systems' are analgous to a shipment of onions which consists of a truck filled with boxes, filled with bags of onions, each onion consisting of several layers, etc.

Your automobile is also a 'system' consisting of several other 'integrated subsystems' (the fuel system, the power train system, the braking system, the ignition system, the cooling system, the lubrication system, the suspension system, the control system, the sound system, etc.) Each of these systems themselves can be comprised of smaller 'systems'.

Some faults/failures can be tolerated without much real impact. For example, a single spark plug can cease to function and it does not collapse the entire 'ignition system'. More than likely your vehicle will still get you where you are going. A 'door system' can fail (hopefully in the closed position) and you still have 3 other doors. A 'windshield wiper system' can fail and, as long as it doesn't rain, you're fine. Some failures can be tolerated for a very short periods (a cooling system failure) or can easily be planned for (do I need to stop at the gas station before I drive 40 miles?)

But other failures within one of the 'embedded systems' can be 'mission critical'. For example, if dirt gets into the fuel line, it can disable 'the fuel system' which in turn disables the 'automobile system' which in turn disables the 'get the kids to school and me to work on time' system. A failure of the braking system can be catastrophic.

In addition, failures in systems external to the 'automobile system' can be just as important. For example, failure of the 'highway system' due to a very heavy snowstorm can also effectively disable the 'automobile system'. Failure of the 'traffic light system' may be either be an annoyance (in a very small town) or catastophic (on a busy highway). And so forth.

This analogy is a bit over-simplified and most of us deal day-to-day with the higher-level systems. But this one of the things that makes the problem so thorny.

I you are a layperson with respect to embedded computer systems, I wouldn't spend too much time worrying about this low-level detail. On the other hand, the information may be useful if you are in some way responsible for identifying, fixing or replacing such systems.

And if you are responsible for your company's PR, don't even think about using this information to justify the absurd claim that "the problem is smaller than we once thought".


-- Arnie Rimmer (arnie_rimmer@usa.net), October 23, 1998.

Yes, Harlan makes a few good points. However, he does play semantics a bit - sure embedded chips are a part of a system, which may or may not be an embedded system, and an embedded system is part of a larger system - thats why it is embedded. And chip interactions are important - replacing a single chip on a board might solve a problem in one case and not in another - but almost no one takes that route - you just replace the entire board and test again.

But as for the semantic issue - there are exceptions to every rule in engineering, I have seen systems that contained a single chip that did everything with nothing else on the board but an oscillator crystal and a capacitor to provide a clock signal. Call it a system if you will - it is, but it is mostly just a chip embedded in a larger analog device that is not - strictly speaking - the usual thing we CS types think of when we talk system. How about your TV or VCR remote? Is it a system? Or a chip with an LED on the end? Or is it part of the TV control system? It all depends on which context you are using at the moment - which all depends on what you are doing at that time. My point is that darn near anything can be called a system if you want to stretch the definition - I never called my hens corn-eating egg-laying manure-producing systems - they are but it is a poor choice of terms. So I don't have nearly the problem Harlan has with terms when talking to non-tech types.

That aside it is a fair article - though I have my doubts as to why he seems to have so much trouble with the figures in the earlier article - if I was asked to critique it I might have complained but I would not have let it slide or guessed what the numbers meant - I would have asked the original source. Also, for completeness, I think there is a rebuttal to this on zdnet somewhere.

However, there are a couple of places where Harlan gets on Mitch for guessing - then he guesses. HOHA. Overall - probably a C+.

-- Paul Davis (davsip1953@yahoo.com), October 23, 1998.

If anyone is going to site the Frautschi paper and lean on it from a technical point of view, I would suggest reading it's rebuttal, which can be found in the archives section of this forum under embedded systems (which is, by the way, what we should be focusing on, NOT embedded chips) It is not semantics when you do a count of the systems at risk vs the chips at risk. It changes the numbers quite a bit.


-- Goldi (goldilucks@yahoo.com), October 23, 1998.

I think you too tightly focused here on some semantics: the actual "system" you want to worry about "survival" at is the plant: the refinery series (cracker, heater, pumps, tower, cooler, cooling water, controls, storage, emissions, and safety blow/relief systems) or the entire sewage plant.

Regardless of whether there is several thousand chips, several tens of thousands, or only two. It doesn't (in the systems-level analysis) matter how many of those there are, nor how many are going to work reliably and correctly.

It only matters whether any are going to fail due to a date calculator. Then, (again, regardless of where the chip is or how it is mounted) the only thing that matters is the result (on the system as a whole) of that failure: Some chips will fail, and nobodyu will notice. Some will fail, people will notice, but the reslut is meaningless. The failure had no actual result.

Some failures (stick open, stick shut, stick without allowing motion) will result in the loss of a system, but nobody cares - or at least not immediately. (For example, if a casino gambling machine quit, would anybody notice that it never paid off? If it paid off every time, the operator (owner) would pull the power supply and void the machine. End of problem.)

The only ones to worry about are those:

Whose failure affects soemthing important,

Whose failure either cannot be replaced or usefully substituted or fixed by hand,

Whose failure "hides" a more critical failure ( the pump sensor fails, the pump doesn't turn on, the sewage storage tank overflows...)

Whose failure can't be repaired - and so it causes more serious problems. A telephone switchboard goes down, and there are no phone lines out of a city or county. A satelite goes off alignment, and the thruster valve sticks open - loses all fuel while spining even faster.

I don't think the number of chips or systems themselves matter - only the number of critical ones not found betweeen now and Jan, 2000.

-- Robert A. Cook, P.E. (cook.r@csaatl.com), October 23, 1998.

Excellent post Robert.

If the power plant continues generating electricity, I am not worried if it takes them a few weeks to get the pollution controls back on line. If the sewage treatment plant is purifying water - I don't care if they are using more anionic flocculants that they really need since they had to go to hand control. In any total system there is a certain tolerance for failures of subsystems - pass the upper limit and the whole system goes down - stay under the limit and the total system still runs, albeit at reduced efficiency.

-- Paul Davis (davisp1953@yahoo.com), October 23, 1998.


What in the world leads you to believe that we will not, in your words "pass the upper limit"? With so many good minds terribly worried that the "upper limit" is well within reach, you seem hell bent that this will not be a big deal. I remember your statements on another thread saying that this may even be GOOD for us, and our economy. Are you just a contrarian? Does God have your ear, with an inside scoop? Somehow the rest of the world does not affect us? Bad news too scary to seriously consider? Inquiring minds want to know.

-- Uncle Deedah (oncebitten@twiceshy.com), October 23, 1998.

I'm still trying to figure this chips vs systems thing. If a system, composed of many (1 or more?) chips, can fail or misbehave due to the failure or misbehavior or ONE of its embedded chips ... then aren't we at least intuitively correct in calling this an "embedded chips" problem? And, indeed, unless one checks the status of every single chip that is part of a "mission critical" embedded system, then it is an open question as to what will happen when 1/1/2000 arrives. (Personally, I hate for stuff like power, clean water, etc., to be open questions....)

-- Jack (jsprat@eld.net), October 26, 1998.

Unca D,

I can't speak for Paul, obviously, but I can speak for me. True, there are a lot of good minds worried silly over Y2K. There are also a lot of good minds who have revised their previously gloomier assessments of the problem. Who is to say which minds are better? Or which assessment is better? I don't really know. (As in, who does.) I know what I think. And you know what you think.

Bad news is not too scary to consider. I subscribed to the whole end-of-the-world thing not too long ago. But I've been happy to see many of those who keep their fingers on the pulse of the problem come out with reports that indicate that the problem as not as bad as previously believed. (No, I don't mean Koskinen's PR fluff on Nightline, either).

Some will revise their estimations of the seriousness of the problem based on new reports; some will not, in the belief that any good news is somehow suspect or untruthful due to some hidden or not-so-hidden agenda. What a person believes is certainly a personal matter.

But what I have maintained, and will continue to maintain, is that for anyone to deny the possibility that better news could be valid, is quite simply a display of ignorance (look up the word in the dictionary -- not necessarily an insult -- ignorant can mean simply, "lacking knowledge"). I'm not saying that the person who holds to a worst-case scenario is an ignorant person; only that the shutting out any better news from consideration due to dogmatism is an ignorant position. And the reverse certainly holds true equally as much.

I don't know if the people who are preparing for food for a year, etc. are right. Those people don't know if I'm right. What I'm saying is that we all need to look at the merits of each others' positions, and weigh them accordingly. To shut out a news report automatically just because it is 'better news', or conversely, 'gloom and doom news', is, yes, ignorant! No matter what one's point of view might be. I've made it pretty obvious what my point of view has come to be in other posts, but I still try to look at all the news, and weigh it accordingly. If a headline says, "End of the World Due to Y2K", I still look at it just as much as a headline that says, "Problem Not as Bad as Previously Believed." We should all do that.

-- John Howard (Greenville, NC) (pcdir@prodigy.net), October 28, 1998.

For those great minds that have stated that it isn't going to be that bad, I have a simple question. What proof do you base your newly acquired perspective on? Do you a files and files full of compliance statements on letterhead, signed by the president of the company, or are you simply conjecturing?

1. Give me (us) the proof.

2. Otherwise convince me (us) why you are taking the stand you are.

Most of us became concerned about the problem, becasue SO many minds were seeing the same ugly future ahead. The primary argument back this past spring was...if you can't get it in writing, then it's not to be believed, and even if you did get it in writing, some couldn't be believed anyway.

So? Where's the proof?

-- Uhuh (notbuying@it.com), October 29, 1998.

Moderation questions? read the FAQ