gnu not unix wrote:
I have wwv via soundcard on the smidge host, it gets to within
100 microseconds (the limit of the soundcard et al), and I've run
a Trimble SV6 and gotten to within single digit microseconds chime.
I also have a Spectracom WWVB clock, and it got down to the
100's of microseconds as well.
Nowhere near wwv around here. Would need an MSF receiver... might well
get one if I can persuade galleon to sell be the bits (they have a
minimum £45 order and the module + ferrite aerial is only £25).
Short of the trick of flying a portable cesium to Boulder, I'm
pretty satisfied that I'm down in the 1's of microseconds to
The One True Tick. My differential drivers and cable delays add
up to perhaps another 100 nanoseconds of offset. With my ADSL
line, the best I've gotten with external stratum 1's has been
a few hundred microseconds offset. My ratio here is on the
order of 5:1 down/up though.
I see something similar with DSL - I'm at 16:1 and it's rare to see the
offset drop below 1.5ish.
It's surprising ntp doesn't take into account network delays in some
way... or even have a config option so I can tune the delay out (a
global time1 parameter?).
See the time-nuts mailing list for Those Others...
time nut. I like that.. :)
Except, that PCI 33mhz gets you. So, your motherboard needs a PCIX or
a 64bit/66mhz slot for the gigabit card.
Unfortunately the machines I have with PCIX & PCI/64 are development
machines as they tend to be the faster ones... maybe in a couple of
generations of hardware upgrades..
Could be.. I put it down to the different processor speed on the boxes.
The smidge host, at 350us, is 400Mhz, and the thrall host, at 400Mhz,
is at 180us. I think its the udp path, myself.
Different network card? Is one DMA capable and the other not, for example?
Also for comparison:
FBSD:
+veracity.mcc.ac 193.63.105.18 2 u 23 64 377 22.989 1.346
0.185
Linux:
-veracity.mcc.ac 193.63.105.18 2 u 3 64 377 22.894 1.062
0.416
So linux seems to have a faster UDP, but on the order of 10us difference
not 200...
Tony
_______________________________________________
timekeepers mailing list
[email protected]
https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
