Re: [gpsd-users] ntpd time accuracy with gpsd and 1PPS

[Greg Troxel]

I recommend  that you read the NTP RFC, and in particular the way four
timestamps are used, and draw some diagrams with differing clocks.  A
key point is that  asymmetric delay and remote clock offsets cannot be
distinguished, because for any given 4 values in the measurement, you
can draw many diagrams.

Basically, with delay total 20ms and offset 0 ms, that can be 10 and 10
(and really 0), or 12 and 8 and being 2 ms slow, or 15 and 5 and being 5
ms slow, etc.  There's no way to figure this out across the net.

pgp6YyOhYUwel.pgp
Description: PGP signature