Am Dienstag, 18. Februar 2014 19:21 hat Richard Cochran [richardcoch...@gmail.com] geschrieben: > On Tue, Feb 18, 2014 at 06:29:11PM +0100, Holzinger, Axel (ALC NetworX GmbH) > wrote > > what is the motivation between the different cases (tuning the clock and > > not tuning the clock inPTP/gPTP). > > Is this demanded by the standard or good practice?
> IEEE 1588 does not tell you how to adjust slave clocks, and gPTP says > *not* to adjust the slave clock. The gPTP people are worried about > "gain peaks" that occur (if I understood correctly) when adjusting a > chain of clocks all at the same time. There are some reports of > simulation results floating around the web that show poor > synchronization in this particular case. Interesting and good to know. Thank you. > > How does another process read the network (PTP) time if the PHC doesn't > > reflect it, because it's free running. > The process must calculate the network time from the offsets. In gPTP > there is a really awful API specified for this, but we offer the > TIME_STATUS_NP management message instead. So this means using the UDS Interface I guess. But does it really make sense to use UDS for reading a precise time. Is it fast enough? I know that also reading the PCH of an 82574L via PCIe for example neither isn't very fast. Good night Axel ------------------------------------------------------------------------------ Managing the Performance of Cloud-Based Applications Take advantage of what the Cloud has to offer - Avoid Common Pitfalls. Read the Whitepaper. http://pubads.g.doubleclick.net/gampad/clk?id=121054471&iu=/4140/ostg.clktrk _______________________________________________ Linuxptp-devel mailing list Linuxptp-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxptp-devel
