On Tue, 28 Jan 1997, Nigel Pearson wrote:

> [Michael R Zucca wrote:]
> > The RTC chip only has a resolution of 1 second. Certain calls in UNIX present
> > time with finer resolution. To emulate that you put a counter on the
> > video interrupt which fires every 60th of a second and you have 60th of
> > a second resolution.

(Actually, it's not the video interrupt, since that fires every 60.15th of
a second, but otherwise the explanation works.)

> 	OK. So, you write a process like xntp which, when system load allows,
> slowly aligns your high resolution clock with the low-resolution RTC chip.

NTP cannot align a clock that is too far out of sync with the external
time source.

I'm not sure why this has become such a hot topic of discussion, since the
changes that I made to the soft interrupt handler code vastly improved the
timekeeping of the clock. I would expect that folks that can use the
`ncrscsi' driver should have much better reliability, given that I have
been seeing a remarkable improvement using the `sbc' driver.

> 	Does anyone know it is possible to detect that a video interrupt
> has been missed?

It is not.  Otherwise, it wouldn't have been missed, right?

--scott