[ros-users] Messaging Performance

Cedric Pradalier cedric.pradalier at mavt.ethz.ch
Tue May 18 21:42:26 UTC 2010


On 05/18/2010 10:05 PM, Ingo Luetkebohle wrote:
> Hi Cedric,
>
> On 05/18/2010 11:00 AM, Cedric Pradalier wrote:
>    
>> A couple of week ago, I've posted a pdf with performance evaluation for
>> inter-node only using local TCP or Sharedmemory. I measured message
>> delivery time, and message frequency, as a function of message size.
>> Search for the "Shared memory image plugin" thread.
>>      
> Thanks for the measurements, I've also been interested in that.
>
> One of the things I noticed on that plot is that the variability is
> fairly large, even without compression (which might be expected to be
> non-deterministic). Is that standard deviation or another statistic?
>
> cheers,
>
>    

These are box plot, so the "whiskers" are the min and max of the data. 
The limit of the boxes are the 1st and 3rd quartile and the bar in the 
middle is the median.

My interpretation is that the peaks comes from the non-real-time 
scheduler in my system (I had other stuff like firefox, thunderbird, and 
so on running on the same machine while testing). Maybe someone could do 
a similar test with Orocos :)

-- 
Cedric
- http://www.asl.ethz.ch/people/cedricp
- http://www.skybotix.com




More information about the ros-users mailing list