About performance of D-Bus

[Thiago Macieira]

On Tuesday 04 November 2008 11:14:50 Jerome Philbert wrote:
> For me, the message speed is important and 4.7ms per message is too slow,
> because :
>  - 100ms is the max reaction time for the user
>  - I have several applications interconnected together through an IPC (will
> be DBus or another)
>  - that can send several messages in the same time
>
> In my system, there are situations where several applications send several
> messages in the same time, and it cannot fit the 100ms.
>
> That is why I would be interested by a quicker direct DBus.

My desktop system (Intel Core 2 Duo @ 2.33 GHz) can do 3270 calls/sec without 
payload, just the roundtrips when both cores are online. That's 0.31 ms per 
round-trip.

With only one core online, the rate drops to 2798 calls/sec.

The system was not idle when I ran those tests. (about 15% of the CPU in use)

In any case, if you have more than 100 messages in the space of 100 ms (i.e., 
more than 1000 calls/sec), then there's something wrong with your design.

Yes, the D-Bus library can be profiled and optimised. No one has done that 
since the last major rewrite, apparently because the developers driving the 
development are on the desktop. The embedded developers should take up that 
task if D-Bus isn't fast enough yet.

-- 
Thiago Macieira - thiago (AT) macieira.info - thiago (AT) kde.org
  Senior Software Engineer - Nokia, Qt Software
  Qt Software is hiring - ask me
      PGP/GPG: 0x6EF45358; fingerprint:
      E067 918B B660 DBD1 105C  966C 33F5 F005 6EF4 5358
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part.
Url : http://lists.freedesktop.org/archives/dbus/attachments/20081104/00cb5abf/attachment.pgp