Re: DBus performance about 20X slower than basic client/server
- From: José Félix Ontañón <felixonta gmail com>
- To: Gnome Accessibility List <gnome-accessibility-list gnome org>
- Subject: Re: DBus performance about 20X slower than basic client/server
- Date: Fri, 16 Jul 2010 02:11:20 +0200
2010/7/15 Bill Cox
<waywardgeek gmail com>
I've done some performance testing of D-Bus vs a simpler IPC message
package. On average, the simplest test message that asks for a
response in DBus takes almost 200 microseconds per call. In the
simpler IPC socket based package, it's only 10 microseconds per round
trip call. The upshot is that DBus is about 20X slower than a basic
socket based client/server communication mechanism.
I'm worried that DBus will become an Orca performance bottleneck. Is
there anything we can do about DBus? The reference implementation we
use is 70K lines, and my vote would be a new implementation designed
for performance.
Bill
_______________________________________________
gnome-accessibility-list mailing list
gnome-accessibility-list gnome org
http://mail.gnome.org/mailman/listinfo/gnome-accessibility-list
Hi Bill,
The D-Bus FAQ[1] told us about how slower it's D-BUS over classic socket basis, and how it can't be recommended for some uses[2] where other IPC mechanism works.
In example, i think a d-bus service emiting a signal for every mouse event it's bad design because d-bus isn't tuned for that. I don't know exactly what orca will expect about d-bus performance.
Cheers!
[1]
http://dbus.freedesktop.org/doc/dbus-faq.html#speed[2]
http://dbus.freedesktop.org/doc/dbus-faq.html#other-ipc
--
http://fontanon.org
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]