| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 457.1 | real time clock = interval clock? | MUNCSS::MIKETTA | Franz W. Miketta - Happy father of a daughter | Thu Apr 30 1987 03:50 | 5 |
| If you mean the interval clock with 'real time clock' then it has
a minimum accuracy of 0.01 percent (the counter is incremented
at microsecond intervals).
Franz
|
| 457.2 | Conference pointer | USHS01::BLANDO | Reality, what a concept! | Thu Apr 30 1987 19:47 | 6 |
| There is an excellent discussion of the VAX architecture specification
for the clock in the VAX conference. (9405::VAX, I don't remember
the node name!) The discussion is old (EI early notes) but still
valid.
FJBlando
|
| 457.3 | 5 minutes a day... | MOSAIC::WASSER | John A. Wasser | Mon May 04 1987 14:18 | 8 |
|
For the Rainbow-100 we decided that an error of 20 minutes per
day was not acceptable but 5 minutes a day was close enough :-)
-John A. Wasser
P.S. What were we supposed to do when the hardware types give us
a 60.0415 Hz clock interrupt?
|
| 457.4 | | CLT::GILBERT | eager like a child | Tue May 05 1987 00:44 | 14 |
| > P.S. What were we supposed to do when the hardware types give us
> a 60.0415 Hz clock interrupt?
Asuming that 60 ticks make a second (as accumulated by the machine),
the 60.0415 Hz clock should cause the machine to gain just under one
minute in a 24-hour period. Not 5 minutes.
VAXen have a NICR register that can be used to adjust the frequency
with which clock interrupts are delivered. Perhaps the Rainbow had
a similar register?
Failing that, you might decide to keep a count-down counter, so that
every 24 ticks, you don't increment the clock by one. This gets
the clock to within one second per day.
|
| 457.5 | EAGLE1::VAX | EAGLE1::DANTOWITZ | David .. BXB1-1/E11 DTN: 293-5356 | Tue May 05 1987 13:34 | 10 |
| -< Conference pointer >-
> There is an excellent discussion of the VAX architecture specification
> for the clock in the VAX conference. (9405::VAX, I don't remember
> the node name!) The discussion is old (EI early notes) but still
> valid.
The node name is EAGLE1.
David
|