Re: [PATCH] HZ=100 assumptions

Eric Lammerts (eric@scintilla.utwente.nl)
Fri, 30 Jul 1999 15:47:07 +0200 (CEST)


On Fri, 30 Jul 1999, Albert D. Cahalan wrote:

> First of all, fix your mail software and/or account information.
> (try the "chfn" command) The linux-kernel archives think you are
> Stephen C. Tweedie, but the email address isn't right for him.

IMHO the linux-kernel archive software is what should be fixed.

> The original PC used a video standard crystal. The divider was set
> to give approximately 65536 ticks per hour, which made it easy to
> detect an hour change. The BIOS needed to update something hourly.

I think that the fact that there are approx. 65536 ticks per hour is a
coincidence, because the divider is set to 65536, the biggest a 8253
timerchip can do. Ralf Brown's interrupt list says:
"there are approximately 18.2 clock ticks per second, 1800B0h per 24 hrs"

> This isn't a very big difference of course. (14.316558 vs. 14.31818)
> DOS would use an "hour" of 3599.592 seconds with the old frequency.
> Linux seems to assume a frequency of 14.31816 -- none of the above.

The frequency is four times the NTSC color subcarrier frequency, which is
defined as (455 / 2) * (525 / 2) * (60 * 1000/1001) Hz =~ 3.579545 MHz.
They used TV crystals because they're cheap.

> Only one way to resolve this: somebody measure a few thousand modern
> computers from different vendors, years, and production runs.

No, the way is resolve this is to ask IBM why they did what they did.
I also have some questions for them about the ISA bus :-).
They're probably too ashamed to answer...

Eric

-- 
Eric Lammerts <eric@scintilla.utwente.nl>
http://www.scintilla.utwente.nl/~eric/

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/