| Re: [chrony-users] gpsd, pps and chrony |
[ Thread Index |
Date Index
| More chrony.tuxfamily.org/chrony-users Archives
]
On Tue, 5 Apr 2011, Miroslav Lichvar wrote:
On Mon, Apr 04, 2011 at 10:02:41AM -0700, Bill Unruh wrote:
Oh, my freetime project at the moment is to try and get the GPIO
interrupts working as a kernel PPS module on the Alix board (hoping to
get higher accuracy than using a serial port...) This coupled with an
old Garmin 18 appears to allow for the possibility of a nice Stratum 1
time server for around the $100 ish mark... Could be cool! Will keep
you informed how it goes...
Cool, I'd like to see how much will the jitter improve with GPIO and
if it'll be better than the Soekris computer
(http://phk.freebsd.dk/soekris/pps/).
You still get
usec level wander (ie it takes about 1-2 usec for the system to service the
interrupt and do the time stamping-- tested by having a program generate an
interrupt by putting out a signal on a parallel pin to the parallel interrupt
line (ACK) timing when that output went out and then having the interrupt
service routine timestamp the incoming interrupt.) but that is pretty good. A
Garmin 18 is not capable of better than that anyway.
Isn't that a delay in the parallel port hw rather than a sw delay? I
think with GPIO the interrupts go straight to the CPU, reducing the
delay significantly.
It might be a parallal port hardware issue, or it might be a software issue.
The cpu has to drop its current calculation and bring in the interrupt
servicing routines, and then read the software clocks. Does that take a 1000
cycles of the cpu? The only way would be to run the tests I guess, using a
nsec timing. of the output and input. As I recall two successive reading of
the clock give about a 1usec difference in time.
---
To unsubscribe email chrony-users-request@xxxxxxxxxxxxxxxxxxxx
with "unsubscribe" in the subject.
For help email chrony-users-request@xxxxxxxxxxxxxxxxxxxx
with "help" in the subject.
Trouble? Email listmaster@xxxxxxxxxxxxxxxxxxxx.