07:57 <rperier> http://postimg.org/image/6t8yzxolb/ <--- Figure 9: Average latency while receiving data, Figure 10: Max latency while receiving data, that's on a zynq zc702

07:59 <rperier> that's amazing how mqueue is fast and has a low latency :/

07:59 <rperier> kdbus still has some improvement to do, because that's not deterministic at all (even if that's interesting compared to the dbus daemon)

09:58 <karlp> what are the units? milliseconds? and what's the x axis?

10:20 <mmind00> rperier: it looks like kdbus will need many other improvements as well: https://lkml.org/lkml/2015/4/27/740

10:23 <mmind00> and https://lkml.org/lkml/2015/4/28/733

10:26 <karlp> oh, is kdbus mean tot be used by dbus? I thought it was a "kernel people wnat a bus too, but need one in kernel space"

10:27 <karlp> can't really understand the desire to use dbus when dbus's own docs say, "integration with this is painful and difficult, you probably want to use more wrapper layers"

10:49 <rperier> karlp: µs

10:49 <rperier> for y and num of packets for x

10:57 <rperier> mmind00: yes this is what I did read on the ML and on lwn :)