Lightening fried my Webtv internet terminal about three weeks ago. I now have a new one and I added a printer. At the time my terminal fried I was developing an astute dissertation, based on the US International Trade Administration report, to post on this forum; so I will give you the short version of my thoughts (lucky you.)

I have thought for some time that one can't make a decision on the assurance of electricity without considering other factors that must be in place before electricity can be generated and delivered. The ITA report is superior in that the data is there to explain our interconnectedness and reliability on the world wide system.

Some of the systems most necessary and, as of now, non-compliant are:

1. Fuel delivery to this country: Saudi Arabia, their pipe lines, water systems, and ports. Venezuela, their electricity, and ports. Non-compliance in Africa (yes, we get oil from there, too.) Approximately 50-60% of our fuel is imported.

2. International Banking Systems: Our largest bank is number twenty-something down the list when compared to international banks. Our banks cannot function unless the international banking system works. The inter-connections are spelled out in detail in the report. When you read this section of the report you will understand the seriousness of this sector. Banking has to work for electric companies to carry on business.

These two sectors alone substantiate my "connect the dots" and "equations" explanations/theories given before on this forum.

I do not believe these dots and equations can be remedied in the time remaining. Ed Yourdon, in his parting letter on his forum, said he believes systems that are not compliant at this time can not be remedied in the time left and can only improve their position by working on and improving contingency plans.

I am glad to be back. I have saved the three weeks of updates from this forum, lots of reading to do. Does anyone know if Factfinder & Co. have come back from Venezuela and Saudi Arabia after fixing problems in these countries? I have read enough of the forum updates to know they are still obsessing over embedded chips. Engineers do tend to obsess. Bless them for the work they do. Marcella

-- Anonymous, June 05, 1999

Answers

Marcella, glad to see you made it through internet deprival! Welcome back. Rick can give you the "Rebecca" character! (See the recent "Can't tell the characters thread...")

On the embedded systems, just so you know things haven't changed too much, the line still most seen in press releases (for oil, gas, whatever) is "We haven't found the embedded chip problems we expected." Which statement, in actuality, does not say *anything* about the status of a company's situation. In order to deduce anything from that kind of statement we'd have to know what percentage of problems they were expecting versus what they've found. Did they expect 15% and only find 10%? Did they expect 5% and only find 1%? Or did they expect 25% and only find 22%? Just because "expectations" were not met, does not necessarily indicate no problems. I've come across one article where percentage data was given about what was found, but we're still usually left with just "less than we expected". See? You didn't miss that much after all!

-- Anonymous, June 05, 1999

Yeah, Bonnie, I read that "Cheers" analogy and absolutley got my feelings hurt that I hadn't caused enough trouble on this forum to rate a character. Rebecca, as I recall, was usually "putting her foot in it" and spending the rest of the time trying to get it out and never quite succeeding. I have stuck my foot in this forum many times trying to get people to put aside the minor issues and move to the crux of the problem(s). I'll take the Rebecca role but only if Rick authorizes it - you know what a hard taskmaster he is. Far be it from me to assume a role he doesn't think I deserve. If he doesn't give it to me my ego is so weak I will never recover. Gosh, it is good to be back. Marcella

-- Anonymous, June 06, 1999

On the embedded systems, just so you know things haven't changed too much, the line still most seen in press releases (for oil, gas, whatever) is "We haven't found the embedded chip problems we expected." Which statement, in actuality, does not say *anything* about the status of a company's situation. In order to deduce anything from that kind of statement we'd have to know what percentage of problems they were expecting versus what they've found. Did they expect 15% and only find 10%? Did they expect 5% and only find 1%? Or did they expect 25% and only find 22%? Just because "expectations" were not met, does not necessarily indicate no problems.

Good heavens. Stop that! Don't you know that critical thinking just confuses people? Just accept such statements at face value without a nanosecond's reflection: you'll feel much better. :-)

-- Anonymous, June 07, 1999

I've mentioned previously that more than 4 billion microprocessors/microcontrollers (MPUs/MCUs) ship each year, but it isn't often you come across the aggregate figure in the press. Usually it's broken down into personal computers, cell phone DSPs, 8-bit MCUs, etc.

But here's a quote from a recent San Jose Mercury News article:

"Tom Starnes, an analyst at Dataquest, a San Jose market research company, said that the number of embedded processors sold in 1998 was 4.8 billion and that only about 120 million of them, about 2.5 percent, had been intended for personal computers."

(orig. New York Times; ref: http://www.mercurycenter.com/premium/business/docs/hiddenchip06.htm)

4.8 billion MCUs in 1998. Another 5 billion in 1999.

Remember these aren't just dumb chips. These are real honest to goodness CPUs. 4-bit, 8-bit, 16-bit, 32-bit, and 64-bit microcomputer controllers. Each and every one of them with some sort of control program or kernel, application code and possible time/date code.

1% of 4.8 billion is 48 million. 0.1% is 4.8 million.

I ask the same question as Bonnie, what is the number that you put on "it's not as bad as we had thought?"

I don't expect an answer, but I'm comfortable with my own off-the-cuff analysis: 1 billion of those chips are 4-bit and 2 billion are 8-bit. It's probable that none of the 4-bit chips and half of the 8-bit chips do no time/date processing at all. That gets rid of 2 billion MCUs (I just cut the problem by more than two thirds, wasn't that generous of me? ;-). of the remaining 2.8 billion MCUs let's say that only 10% of the designs use time/date functions: 280 million MCUs with potential Y2K problems shipped last year. Let's say only 5% of them actually had Y2K issues: 14 million MCUs with problems shipped in 1998. At least 10% of designs will never be corrected as it is not economically feasible to do so (I honestly expect this to be more like 30% but history will tell): 1.4 million MCUs shipped in 1998 with known Y2K problems. I expect the pattern for 1999 to not be much different. So let's round it down to 1.2 million and call it 100 thousand MCUs with known Y2K problems ship each and every month, right up to December 1999 and on into 2000.

Let's say I'm off by an order of magnitude (a factor of 10) in this little back of the napkin analysis. That would bring the number down to 10 thousand units per month best case.

Conclusion: in December 1999 companies around the world will ship a quantity of products containing at least 10 thousand embedded MCUs with known (or ignored) Y2K problems.

This is being done knowingly because it is simply not economically feasible to upgrade, replace and/or repair all the billions and billions of little computer brains out there.

Maybe all the problems will be benign. Maybe they won't. History will tell.

"207 days and counting" (in the voice of the announcer on the off-shore oil rig from the James Bond movie "Diamonds are Forever" :-)

--aj

-- Anonymous, June 07, 1999

AJ, As usual I find your analysis extremely interesting and challenging. I like your logic, math, but disagree with the sources that provide the basis for the assumptions and numbers you use. I do not have time to do the additional research needed to discuss this at length, but hope to over the next few weeks. You have done your homework, and I need to do mine to do the topic justice.

Regards,

-- Anonymous, June 10, 1999

A small addition and correction to the 4.8 billion MCUs number.

Digital Signal Processors (DSPs) are not included in that number. However, DSPs rarely, if ever, do any time-of-day time/date processing, so it's a moot point.

Also, not included in the 4.8 billion figure are Application Specific Integrated Circuits (ASICs) that have an MCU included as part of their function. Many ASICs these days include an 8-bit ot 16-bit MCU as part of their functionality but they do _not_ get counted as MCUs/MPUs by market research companies such as DataQuest. These ASIC parts are perfectly capable of having time/date functionality.

Another new class of MCU/MPU is the "system on a chip", similar to an ASIC but with a different moniker. I'm not sure if they're counted or not.

Also, many military and security agencies have their own fabs (chip plants), or as a minimum the chips and numbers are classified.

The bottom line is that the 4.8 billion figure for the number of MCUs/MPUs produced each year is already knowingly conservative.

-- Anonymous, June 10, 1999