Soft Landing for Embedded Chips?

Few Y2K Problems Expected, says Gartner

By Thomas Hoffman 08/30/99 Embedded chips in oil rigs, power plants and pacemakers might not suffer as many year 2000-related glitches as some experts once feared.

According to new research published earlier this month by Gartner Group Inc. in Stamford, Conn., only a "small percentage" of microprocessors and microcontrollers will experience any date-related problems. And even if they do record a date inaccurately, they won't necessarily shut down altogether.

For example, some microprocessors act as programmable logic controllers (PLC) that were designed to perform such tasks as opening a valve in a heating unit. If the PLC does misinterpret the year 2000 rollover, it might still open the valve but stamp the date incorrectly.

Besides, microprocessors that don't have real-time clocks or aren't connected to devices with real-time clocks "cannot generate a date from thin air," according to the Gartner study (www.gartner.com/worldy2k).

And even if a microprocessor is tied to a real-time clock, it may lead to "anomalous processing" but not necessarily an outright failure, said Gartner analyst Lou Marcoccio.

Still, one class of embedded systems, known as Large-Scale Embedded Systems (LSES), appears to be more date-sensitive than less-complex microcontrollers and microprocessors. LSESs are often PCs used to control factory floors or heating and air conditioning systems.

User testing appears to support Gartner's research. For example, General Motors Corp. has tested 1.4 million devices at its 150 factories and found that it had to make changes in fewer than 15% of its embedded systems, said John Ahearne, GM's communications manager for the year 2000 project.

Computerworld

******************************************************************* This kind of reporting just pisses me off, especially considering the fact that it comes from respectable computer industry publication. In their first report for the 4th Quarter of 1998 (which they now sell for $200) they claimed that only about %.001 of embedded systems will fail. They modified their assessment as of the latest report which is mentioned in this article. Kudos to Ford for being able to check 1.4 million embedded systems (being #2 in Fortune 500 helps), but the fact that they found MORE THAN 14% of the embedded systems to be noncompliant tells me that Gartner is incorrect (to put it mildly), and scares the living hell out of me, because not many companies have money and ability to check all their systems.

-- Boris Ushumirskiy (MSIS@cyberdude.com), September 01, 1999

Answers

Shhhhh! Don't tell anybody but the dam is about to burst! If we tell people they'll panic and try to escape.

-- cody (cody@y2ksurvive.com), September 01, 1999.

Boris ---

Thought the same thing. How these "giants of industry" can pass "less than 15%" off as good news is absolutely amazing. The world is insane.

I still can't believe my eyes. If this is really ANYWHERE close (say within 1/15th) to average we are in deeper trouble than imagined, especially oil/gas.

-- Jon Johnson (narnia4@usa.net), September 01, 1999.

Besides the inference of extrapolating out the "less than 15%" to be noncompliant as being good news over all for all embedded systems, did you notice all the other "conditioners" such as "might not" and "won't necessarily"? "Sheesh" to quote Flint.

-- Valkyrie (Anon@please.net), September 01, 1999.

... fewer than 15% ...

Fewer than 15% ?!!

FEWER than 15% !!!

oh - I feel MUCH better now.

-- Dan G (holyshit@thats.bad), September 01, 1999.

Pardon my error, but after I saw "less then 15%" I forgot everything else. Kudos have to be given to GM - #1 Fortune 500 (not to Ford). Again, I am sorry.

-- Boris Ushumirskiy (MSIS@cyberdude.com), September 01, 1999.

Boris,

You have a url?

-- WAtcher7 (anon@anon.com), September 01, 1999.

http://www.computerworld.com/home/print.nsf/all/990830BDA6

-- Boris (MSIS@cyberdude.com), September 01, 1999.

Oh, OK, spend whatever it takes to prevent logging a date with an '0' rather than an '00'. I just saw a bundle of your tax dollars go by to correct EXACTLY that. Well spent. (NOT!)

But I really DON"T care if the power company clerk has to put up with '0' instead of '00' as long as the power keeps flowing.

-- Paul Davis (davisp1953@yahoo.com), September 01, 1999.

You are either stupid or simply ignorant. Date calculations inside the embedded systems have nothing to do with clerk seeing one or two zeroes. I would suggest reading more about it before ever posting on this board.

-- Boris (MSIS@cyberdude.com), September 02, 1999.