[time-nuts] ``direct'' RS-232 vs. RS-232 via USB vs. PPSdecoding cards
hmurray at megapathdsl.net
Wed Feb 22 03:49:24 EST 2017
qb4 at comcast.net said:
> I found while looking at the datasheets for newer Intel server ethernet
> cards that they have the ability to timestamp GPIO pin transitions, but none
> of them have their internal timebase synchronized to a counter in the CPU.
> It looks like they are clocked from a separate XO on the card. Maybe if it
> was synchronized to the PCIe clock / BCLK they could take advantage of that
> new Always Running Timer in Skylake processors. I'm surprised that Intel
> hasn't made a big deal about it. Support for it was added in the Linux
> e1000e driver early last year.
I think that's enough to synchronize the counter on the Ethernet card with
the CPU counter. You need either something like a PPS connected to both the
Ethernet and some pin on the CPU, or you need some pin the CPU can control
connected over to the Ethernet.
Do you have any info on the Linux support? What magic term to I search for?
(I'm guessing it's a magic ioctl.)
These are my opinions. I hate spam.
More information about the time-nuts