On Tue, Aug 16, 2016 at 3:36 PM, Attila Kinali <[email protected]> wrote:
> I disagree here. With the APU and its IRQ system you get already to > sub-us PPS time stamping (probably in the 10ns-100ns range). > An APU1 post said they had 50ms of jitter (as measured by Chrony) so they assumed the PPS driver was broken on the hardware. Pascal suggested elsewhere that the combined LPC plus system IRQ jitter would be too much for NTP. While I don't believe this I'm disinclined to test it . So I was -- clearly too briefly -- implying that there are solutions that should be less expensive* and known to achieve micro-second jitter. I had no reason to suspect Pascal would overestimate the issues however another APU1 owner** used an mPCIe serial card and an mPCIe GPS card and got expected microsecond jitter (as measured by NTP) results. Again this is not quite what the OP is proposing but it's encouraging. These are time sync. protocol results and don't speak to sub-microsecond time-stamping in the hardware. * which is not to say pcengine hardware is expensive. It's not. ** https://wiki.polaire.nl/doku.php?id=versalogic_vl-mpeu-g2_ntp _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
