[time-nuts] RPi NTP was Net4501's cheap...
tic-toc at bodosom.net
Sat May 25 12:47:38 EDT 2013
On Sat, May 25, 2013 at 2:43 AM, David J Taylor wrote:
> Just what devices were you hot-plugging to produce this problem? One with
> an initial current surge outside the USB spec I could understand.
I was plugging in a USB to RS232 adapter. If you read the various bits
you'll find about the USB on Rev.B boards you'll see all the work-arounds
suggested here as well as an implication that Raspberry intends to fix the
problem in a later release.
> NTP performance on the three Raspberry Pi cards here can be of the same
> order as FreeBSD running on an Intel Atom PC:
Yes, I've looked at your stats. As I said earlier you put me onto the Pi
and the Sure boards. The offsets on the Pi are fine. I just don't like the
"weak" network performance which you can see in the ntpq output for the
Pi. Since I'm still testing bits all four of my clocks are on the same
gigabit switch so I don't expect to see simultaneous negative and positive
offsets. The Laureline is jittery too but at least it always presents the
same offset "polarity". I'll see how the Beaglebones do, check out the
next rev. of the Laureline and then dither some more.
More information about the time-nuts