Hi Richard, Thanks for your inputs. They were of great help. When we ran ptp4l with hardware timestamping, without synchronization of the system clock with the hardware clock on the master node, we are able to get around 25 nanoseconds offset between the master and slave clock which seems to be a good and desired value.
So as per our understanding, we were getting 1000 ns accuracy in the previous test because our system clock was unstable and we were synching it with the hardware clock on master (because we do not have a grandmaster yet). *Here are logs for the recent test we performed based on your inputs:* ptp4l[233.770]: selected /dev/ptp0 as PTP clock ptp4l[233.771]: port 1: INITIALIZING to LISTENING on INIT_COMPLETE ptp4l[233.772]: port 0: INITIALIZING to LISTENING on INIT_COMPLETE ptp4l[234.902]: port 1: new foreign master d4bed9.fffe.318e3e-1 ptp4l[238.903]: selected best master clock d4bed9.fffe.318e3e ptp4l[238.904]: port 1: LISTENING to UNCALIBRATED on RS_SLAVE ptp4l[240.907]: master offset 277128084 s0 freq +0 path delay -29250 ptp4l[241.908]: master offset 277208494 s1 freq +80394 path delay -29250 ptp4l[242.908]: master offset -8 s2 freq +80386 path delay -29250 ptp4l[242.908]: port 1: UNCALIBRATED to SLAVE on MASTER_CLOCK_SELECTED ptp4l[243.908]: master offset -15434 s2 freq +64958 path delay -13842 ptp4l[244.908]: master offset -11321 s2 freq +64441 path delay -2503 ptp4l[245.908]: master offset 4624 s2 freq +76989 path delay -2503 ptp4l[246.908]: master offset 7188 s2 freq +80940 path delay -1630 ptp4l[247.908]: master offset 5105 s2 freq +81014 path delay -137 ptp4l[248.909]: master offset 4476 s2 freq +81916 path delay -137 ptp4l[249.909]: master offset 1461 s2 freq +80244 path delay 1355 ptp4l[250.909]: master offset 1558 s2 freq +80779 path delay 1424 ptp4l[251.909]: master offset 1130 s2 freq +80819 path delay 1494 ptp4l[252.909]: master offset 681 s2 freq +80709 path delay 1494 ptp4l[253.910]: master offset 315 s2 freq +80547 path delay 1530 ptp4l[254.910]: master offset 168 s2 freq +80495 path delay 1565 ptp4l[255.910]: master offset 112 s2 freq +80489 path delay 1568 ptp4l[256.910]: master offset 58 s2 freq +80469 path delay 1577 ptp4l[257.910]: master offset -44 s2 freq +80384 path delay 1598 ptp4l[258.911]: master offset -31 s2 freq +80384 path delay 1598 ptp4l[259.911]: master offset -12 s2 freq +80394 path delay 1586 ptp4l[260.911]: master offset -23 s2 freq +80379 path delay 1577 ptp4l[261.911]: master offset 73 s2 freq +80468 path delay 1577 ptp4l[262.911]: master offset 20 s2 freq +80437 path delay 1584 ptp4l[263.912]: master offset 63 s2 freq +80486 path delay 1577 ptp4l[264.912]: master offset 17 s2 freq +80459 path delay 1577 ptp4l[265.912]: master offset -35 s2 freq +80412 path delay 1582 ptp4l[266.912]: master offset -24 s2 freq +80412 path delay 1582 ptp4l[267.912]: master offset 25 s2 freq +80454 path delay 1577 ptp4l[268.912]: master offset -24 s2 freq +80413 path delay 1577 ptp4l[269.913]: master offset -19 s2 freq +80411 path delay 1577 ptp4l[270.913]: master offset 33 s2 freq +80457 path delay 1570 ptp4l[271.913]: master offset -66 s2 freq +80368 path delay 1584 ptp4l[272.913]: master offset 36 s2 freq +80450 path delay 1584 ptp4l[273.914]: master offset 13 s2 freq +80438 path delay 1581 ptp4l[274.914]: master offset 29 s2 freq +80458 path delay 1581 ptp4l[275.914]: master offset -23 s2 freq +80414 path delay 1584 ptp4l[276.914]: master offset -11 s2 freq +80419 path delay 1584 ptp4l[277.914]: master offset 3 s2 freq +80430 path delay 1584 ptp4l[278.914]: master offset 44 s2 freq +80472 path delay 1584 ptp4l[279.915]: master offset -12 s2 freq +80429 path delay 1582 ptp4l[280.915]: master offset 4 s2 freq +80442 path delay 1582 ptp4l[281.915]: master offset 21 s2 freq +80460 path delay 1577 ptp4l[282.915]: master offset 8 s2 freq +80453 path delay 1572 ptp4l[283.915]: master offset -40 s2 freq +80408 path delay 1572 ptp4l[284.915]: master offset 4 s2 freq +80440 path delay 1570 ptp4l[285.915]: master offset 23 s2 freq +80460 path delay 1566 ptp4l[286.916]: master offset -24 s2 freq +80420 path delay 1566 ptp4l[287.916]: master offset 24 s2 freq +80460 path delay 1565 ptp4l[288.916]: master offset -21 s2 freq +80423 path delay 1565 ptp4l[289.916]: master offset 25 s2 freq +80462 path delay 1566 ptp4l[290.916]: master offset 10 s2 freq +80455 path delay 1566 ptp4l[291.916]: master offset -70 s2 freq +80378 path delay 1569 ptp4l[292.916]: master offset 7 s2 freq +80434 path delay 1566 ptp4l[293.916]: master offset 52 s2 freq +80481 path delay 1566 ptp4l[294.917]: master offset 5 s2 freq +80450 path delay 1566 ptp4l[295.917]: master offset -15 s2 freq +80431 path delay 1567 ptp4l[296.917]: master offset 27 s2 freq +80469 path delay 1567 ptp4l[297.917]: master offset -36 s2 freq +80414 path delay 1579 ptp4l[298.917]: master offset -29 s2 freq +80410 path delay 1579 ptp4l[299.918]: master offset 11 s2 freq +80441 path delay 1579 ptp4l[300.918]: master offset 33 s2 freq +80466 path delay 1573 ptp4l[301.918]: master offset -11 s2 freq +80432 path delay 1568 ptp4l[302.918]: master offset 36 s2 freq +80476 path delay 1565 ptp4l[303.918]: master offset 16 s2 freq +80467 path delay 1565 ptp4l[304.918]: master offset -39 s2 freq +80417 path delay 1568 ptp4l[305.919]: master offset 3 s2 freq +80447 path delay 1568 ptp4l[306.919]: master offset -14 s2 freq +80431 path delay 1564 ptp4l[307.919]: master offset 27 s2 freq +80468 path delay 1563 On Thu, Oct 25, 2018 at 5:41 PM Dolly Gyanchandani < dollygyanchandani1...@gmail.com> wrote: > Hi Miroslav, > > Thanks for your response. Please find my comments inline below. > > The accuracy of which clock would you like to measure? The system clock or > the PTP clock on the NIC? > > We want to measure the accuracy of the PTP clock now. But eventually, we > would need to measure the accuracy of the system clock as well. Could you > suggest reliable ways to measure both? > > A separate and more accurate time source is necessary. For measuring > the accuracy of SW timestamping you can use HW timestamping. > > What do you mean by this? Do we need to have PTP enabled NICs to be able > to test software timestamping? Not sure if we understand this point > correctly. Could you elaborate a bit? > > Thanks, > Dolly > > On Thu, Oct 25, 2018 at 12:23 AM Richard Cochran <richardcoch...@gmail.com> > wrote: > >> On Wed, Oct 24, 2018 at 02:14:44PM +0530, Dolly Gyanchandani wrote: >> > Our current measure of accuracy is based on* master-slave offsets >> reported >> > by the ptp4l process.* >> >> Log output, please. >> >> > *In Hardware Timestamping, *we are getting* >1000 nanoseconds* offset >> from >> > the master which is far larger than the *expected value of around 50 >> ns* of >> > PTP H/W Timestamping. >> >> That is really bad. >> >> > (Since we do not have a grand-master currently, we are using Master's >> > system clock as the Master clock for phc2sys) >> >> First try leaving the PHC time on the NIC free running. >> >> Thanks, >> Richard >> >
_______________________________________________ Linuxptp-users mailing list Linuxptp-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxptp-users
