I've installed ntp-4.2.2p3-RC1 on one of my Sun Ultra 10 workstations"
sunburn_$ uname -a
SunOS sunburn 5.10 Generic_118822-25 sun4u sparc SUNW,Ultra-5_10

I built with the Sun Studio 11 compiler.

It seems to keep time far less well than the ntp-4.2.0 that it replaced.
sunburn_$ ntptime
ntp_gettime() returns code 0 (OK)
time c8764712.42c1e000 Sat, Jul 29 2006 16:44:02.260, (.260771),
maximum error 282734 us, estimated error 11542 us
ntp_adjtime() returns code 0 (OK)
modes 0x0 (),
offset 0.000 us, frequency 11.852 ppm, interval 4 s,
maximum error 282734 us, estimated error 11542 us,
status 0x1 (PLL),
time constant 6, precision 1.000 us, tolerance 512 ppm,
pps frequency 0.000 ppm, stability 512.000 ppm, jitter 200.000 us,
intervals 0, jitter exceeded 0, stability exceeded 0, errors 0.

unburn_$ ntpq -p
remote refid st t when poll reach delay offset
================================================== ============================
*sunblok .GPS. 1 u 533 1024 377 0.482 -10.894

The error/offset is consistently between -10 and -20 milliseconds while
synchronized with another Sun Ultra 10 on my LAN. That system is
synchronized by a Motorola Oncore M12+T GPS receiver. Other systems on
the LAN seem to be able to keep synchronization within less than 1
millisecond. The jitter seems quite high.

At this point I don't know if the problem is Solaris 10, or NTP 4.2.2p3-RC1.

Anyone have any thoughts/suggestions?