> On Jan 10, 2017, at 03:21, Miroslav Lichvar <mlich...@redhat.com> wrote: > > On Mon, Jan 09, 2017 at 09:02:14AM -0800, Denny Page wrote: >> The program up if you want to have a look. I am still working on the >> guidelines for use. >> >> https://github.com/dennypage/ethtscal >> >> My testing has been principally focused on the i354. I will also be testing >> the i211, which I believe should be very close to the i210, but I need to >> finish the i354 first. Unfortunately it’s a long iterative process to >> balance the tx and rx values. > > This is very interesting and I'm curious to see how you balance the > tx and rx values. If I know the expected delay is x and I measure > delay y, how do I distribute the (x-y) difference to the tx and rx > comp values? Is there some symmetry in the RX and TX timestamping > error?
Well, that’s the bad part—without a scope or other interval hardware device the only way to determine the values is by trial and error. The short version is that you establish a baseline for different speeds using loopback cables, and then work the values with different speeds through a switch until all the values fit. A long and painful process. And it doesn’t help that switches change latency between 10/100Mb and 1Gb. The tx and rx compensation values are asymmetric on the Intel chips I’ve tested. Between 1:2.1 and 1:2.6 depending upon speed. Denny -- To unsubscribe email chrony-dev-requ...@chrony.tuxfamily.org with "unsubscribe" in the subject. For help email chrony-dev-requ...@chrony.tuxfamily.org with "help" in the subject. Trouble? Email listmas...@chrony.tuxfamily.org.
