Well, I finally got my hands on a BIOS dump from a unit that exhibits TD. I can tell you that this BIOS has no special date-specific code, and runs no risk of violating the 244us window, although it does run the risk of returning bad times or dates at ANY time, not just after 2000. It runs this risk because it checks for update-in-progress between reads of each register, so it can read one register before update, and the next after, and when and if this happens, the time/date is guaranteed to be off by at least one second if update happens between reads. If this is the last second of a minute, hour, day, month or year, it will be incorrect by that amount. This possibility is therefore not a y2k problem at all.

I suspected that DOS might read the RTC directly, but I have been assured by people much more knowledgeable of DOS than I am that DOS *never* reads the RTC directly.

This leads to a conundrum: If TD on this unit is not caused by the BIOS, and isn't caused by DOS, and nobody else is involved, and changing the RTC to a buffered unit fixes the problem (which it does), then WHAT'S HAPPENING HERE? Beats me.

-- Flint (flintc@mindspring.com), November 15, 1998

Answers

That is the craziest thing I have ever heard. But one thing I have noticed in the units I have inspected for Y2K problems - was the unit really cheap when it was new? A cheap unit seems likely to have more problems than an expensive one - you get what you pay for I guess. I don't guess you want to tell who the BIOS mfr is for legal purposes, but was it in the US or offshore?

-- Paul Davis (davisp1953@yahoo.com), November 15, 1998.

Flint,

You may want to e-mail North with what you found out. He is doing a victory dance thinking that this has been proven, but yet he makes no mention it is a anytime problem.

Rick

-- Rick Tansun (ricktansun@hotmail.com), November 15, 1998.

Can you please cut the geek-speak and tell us what you are talking about in English? The only TD I know of is a Touchdown!

-- Mary (mary@lost.com), November 15, 1998.

Mary

I wrote a long explanation and sent it to your bogus mailing address. It bounced back. Aren't you glad you're so secretive? I guess you'll just have to do the research on your own.

-- Flint (flintc@mindspring.com), November 15, 1998.

Mary, and others not familiar with "TD" in this context:

What's being discussed here is a phenomenon first observed last year by a few Y2K testers who were running computers with the date set later than 1999. Initially, this phenomenon was called "Time Dilation", abbreviated to "TD", because those who first observed it thought that the symptoms they observed were caused by their computers' clocks running faster than normal. For example, over a two-week test interval, the dates recorded by the system advanced from January 2000 to December 2000, about twenty times the normal rate of advance.

Now, it seems well-established that "time dilation" is a misnomer because it is NOT a result of clocks running faster than normal. (E.g., in computers exhibiting this phenomenon, the system date moves in abrupt jumps, sometimes even backwards.)

It now appears that this phenomenon affects only computers with certain types of internal components. Its relationship to Y2K is only coincidental; "TD" is a different type of problem, but it just happens to show up much more often when the date is 2000 or later on a computer which exhibits "TD" than when the date is earlier than 2000.

Many folks now call the phenomenon "Time/Date Jump" or "Crouch-Echlin Effect" (Crouch and Echlin were two principal investigators of it).

The article "Time Dilation - Where Are We Now", written by Harlan Smith, at http://www.y2ktimebomb.com/Computech/Issues/hsmith9845.htm has more explanation. Though it's mostly a techie article, there is some fairly plain explanation within it.

If you want to see a historical record of what the early observers found and thought about it, see http://www.nethawk.com/~jcrouch/td01.txt

-- No Spam Please (anon@ymous.com), November 16, 1998.

No Spam Responder,

THANK YOU very much for sharing that information. It is very interesting and much easier to understand! I'm glad someone understands the meaning of the word FORUM.

Thanks, Mary

-- Mary (mary@enlightened.com), November 16, 1998.

Yup - forum is one more than threeum.

To the Flint person: any indication (given the confusion and difficulty of this research), of a TD effoect in embedded controller?

Reviewing the refinery and chemical process industry trade magazines, there are hundreds, if not thousands, of electronic sensors (temperature, pressure, differential pressure, flow, weight, fiuld level, controllers, regulators, feedback position indicators, etc.) in a typical refinery.

Wonder if Chevron is checking those too?

-- Robert A. Cook, P.E. (Kennesaw, GA) (cook.r@csaatl.com), November 16, 1998.

Robert,

As has so often happened in the English language, "forum" has at least a dual etymology. Perhaps Mary was referring to its being the antonym of "aginum". :-)

-- No Spam Please (anon@ymous.com), November 18, 1998.