Need help about the performance of RTW/RTAI/Comedi

[Paolo Mantegazza]

wicky.zhang wrote:
> 
> ----- Original Message ----- From: "Roberto Bucher" 
> <roberto.bucher at supsi.ch>
> To: "wicky.zhang" <wicky.zhang at gmail.com>
> Sent: Thursday, January 19, 2006 1:49 PM
> Subject: Re: Need help about the performance of RTW/RTAI/Comedi
> 
> 
>> Hi Wicky
>>
>> we haven't reached such sampling frequencies till now, thus I can't say
>> exactly what the problem is. First, you have to consider that a i386
>> machine
>> is not a DSP, you have ltencies which are greater than the DSP latencies.
>> Paolo in Milano reaches sampling frequencies of about 40Khz, for special
>> applications.
>>
>> Now, to your problem. I think that when you don't open the xrtailab
>> application, the system works. Can you confirm it?
> 

I think it is even 45, 4 channels, but they use COMEDY async mode, i.e. 
acquisition timed and executed independently and data acquired when all 
available. Such a mode is not implemented in RTAILab, at least for the 
RTW I think the related S-fuctions use COMEDI data read/write.

Paolo.
> 
> Yes, i have used oscillograph and signal generator to observe the input and
> output of my DAQ card.
> 
>>
>> Now, what happens if you run "xrtailab" on a second PC (with Linux RTAI
>> installed, of course)? Do you have the same behaviour?
> 
> 
> It will die when i run "xrtailab" on a second pc. My system have a 
> target PC
> with a remote PC for saving the resource of the target PC.
> 
>>
>> If you don't have a second PC with Linx RTAI, you can try to install
>> RTAI-XML
>> ( http://artist.dsi.unifi.it/rtaixml/ ) and monitor the RT task from a 
>> web
>> browser.
> 
> 
> Oh? it is a new idea. i have never notice it.
> 
> Thank you!
> 
> Best regards
> 
> Wicky
> 
>>
>> Regards
>>
>> Roberto
>>
>> On Thursday 19 January 2006 00:36, you wrote:
>>
>>> Hi Roberto,
>>>
>>> Appologize for disturbing you again.
>>>
>>> After trying many times and referring many documents, I can do nothing
>>> about how to improve the performance of my simplest test demo (output
>>> path
>>> = a sine wave generator--->gain--->Comedi data write, while input path =
>>> Comedi data read--->sfun_rtai_scope). The demo is based on
>>> Matlab/RTW7.1(windows version)+RTAI3.2+Comedi0.7.70.  I tried NI6024E 
>>> and
>>> PCI9118dg of ADLINK both, the following is my test result:
>>>
>>> 1: For 1 channel A/D,  the sample rate can be set up to 30KHz. But when
>>> raised to 50KHz or higher, once using remote RTAI-LAB to connect the
>>> target
>>> system, the computer will not response the connection, even not response
>>> to
>>> either mouse or keyboard. The only thing i can do is to push the reset
>>> button.
>>>
>>> 2: For 6 channels A/D, the sample rate can be set up to 5KHz only. when
>>> raise it, the same thing happens.
>>>
>>> I have been suggested to check the cmd->flags|=TRIG_WAKE_EOS, which 
>>> would
>>> cause interrupt every scan, but i don't know how to set or clear it in
>>> the
>>> auto-generated code, even after reading many of the corresponding
>>> documents, including rtai, comedi and Matlab/RTW.
>>>
>>> I am wondering that whether the performance can be raised with the
>>> auto-generated code? or should there be some tricks to improve it? I 
>>> hope
>>> my system can reach 10KHz sample rate with 6 channels A/D at least, can
>>> you
>>> give me some hints? thank you!
>>>
>>> Best Regards,
>>>
>>> Wicky
>>
>>
>> -- 
>> University of Applied Sciences of Southern Switzerland
>> Dept. Innovative Technologies
>> CH-6928 Lugano-Manno
>> http://www.dti.supsi.ch/~bucher
> 
> 
> 
> _______________________________________________
> RTAI mailing list
> RTAI at rtai.org
> https://mail.rtai.org/cgi-bin/mailman/listinfo/rtai
>