synchronizing sound cards in a cluster
josip at nianet.org
Fri Mar 14 07:02:16 PST 2003
Robert G. Brown wrote:
> a) Check out the documentation on http://www.ntp.org/documentation.html.
> From what it says, you can synchronize at roughly the level of network
> latency with ntp alone, so you can (I would expect) get an otherwise
> quiet LAN sync'd to a millisecond or even less. NTP does correction
> over a long time and damps to a common clock, so you MIGHT get down
> below the 1 ms mark over time. I doubt that ntp alone would make 10
NTP is great; I've been using it for 15 years. Back then, our HP-UX machines
could reliably synchronize local clocks to sub-msec levels using NTP.
Our Linux cluster clocks are synchronized to our NTP broadcast head node
within about 1-5 msec (estimated based on clockdiff and ntpq data). If you
need higher precision, NTP polling can do even better. Although network
latency within a cluster can be almost 100 usec, this is repeatable and over
long time 10 usec synchronization with the local NTP polling server is feasible.
While NTP can adjust local clocks very accurately, Linux system calls which
read the clock are another source of timing errors. Although gettimeofday()
returns microseconds, this is not always reliable down to the last digit. On
the average, you can probably expect 10 usec or better results. For example,
on our machines, ntpq reports local clock jitter of 8 usec.
Dr. Josip Loncaric, Research Fellow mailto:josip at nianet.org
National Institute of Aerospace http://research.nianet.org/~josip/
144 Research Drive mailto:j.loncaric at larc.nasa.gov
Hampton, VA 23666, USA Tel. +1-757-766-1395 Fax +1-757-766-1401
More information about the Beowulf