[LAU] yoshimi creeps... 0.051

[rosea grammostola]

Folderol wrote:
> On Sat, 16 Jan 2010 13:34:45 +0100
> rosea grammostola <rosea.grammostola at gmail.com> wrote:
>
>   
>> Folderol wrote:
>>     
>>> On Sat, 16 Jan 2010 13:07:23 +0100
>>> rosea grammostola <rosea.grammostola at gmail.com> wrote:
>>>
>>>   
>>>       
>>>> cal wrote:
>>>>     
>>>>         
>>>>> The main change in 0.051 is to the recorder start logic, which sort of goes like -
>>>>> - Select your wav file, and the recorder becomes armed.
>>>>> - When armed, recording starts with the first note-on event, or a click on
>>>>>   the Record button.
>>>>> - When armed, a click on the Stop button disarms the recorder.
>>>>> - When recording, a click on the Stop button stops the recorder, leaving it
>>>>>   armed.
>>>>>
>>>>> <http://sourceforge.net/projects/yoshimi/files/>
>>>>>
>>>>> Bon apetit.
>>>>>   
>>>>>       
>>>>>           
>>>> Yoshimi is not usable for me as an instrument because of the many xruns.
>>>>
>>>>
>>>> \r
>>>>     
>>>>         
>>> I'm astonished that you have this problem. I have found it to be far
>>> less prone to these than the zyn 2.4 source.
>>>   
>>>       
>> What is less prone? Having regular xruns but less then it was before? Do 
>> you get xruns? More when using certain instruments? Or playing chords?
>>     
>
> With yoshimi I get no xruns at all while playing 3-4 note chords with
> the left hand plus a melody line with the right hand. This is true even
> with my most complex patches. With the most recent versions, I only see
> xruns while changing instrument patch at the same time as actually
> playing notes (which is a pretty daft thing to do).
>
>   
>> Especially when I switch instruments and play chords, is quite simple 
>> sometimes to reach to 1000 xruns in a few minuts... I think for an 
>> instrument you should have almost none xruns in an hour and strive to 
>> get none...
>>     
>>> What is your setup?
>>>
>>>   
>>>       
>> Debian testing, Jack2, RT kernel :
>> Linux debian 2.6.31.6-rt19-1-686 #1 SMP PREEMPT RT Thu Nov 12 13:22:20 
>> UTC 2009 i686 GNU/Linux
>>     
>
> This is far more recent that what I'm working with (2.6.29) - I know
> there have been problems with some rt kernels but I don't know which
> ones. Also I tend to work with targeted music distros, such as 64studio,
> where some extra optimisations are done.
>
>   
>> @audio - rtprio 99
>> @audio - memlock unlimited
>>     
>
> This looks normal, but I don't see a 'nice' setting. Mine is -10
>  
>   
>> I'm using an maudio keystation usb midi keyboard
>>     
>
> I don't know this particular keyboard but I have heard that some USB
> ones have problems. I think this actually a mobo problem where there
> are shared interrupts, but I'm far from certain about that.
>
>   
>> I think I have a pretty proper configured system... No problems with 
>> other apps.
>>     
>
> Are the other apps synths or sample players? and do you drive them
> directly from the keyboard?
>
> P.S.
> Just done some test and I can create xruns if I use my most processor
> intensive patches 'hyper matrix' and 'master synth low', but to do
> this I have to play quite fast fully chorded, which is really not what
> these patches are intended for. Both of these are four instrument voice
> 'kits' with long release times and significant extra effects.
>
> Are you running similar heavy-duty voices?
>
>   
Mhh it sounds like your yoshimi with your setup is performing much 
better then mine.

Maybe it's the hardware? I've just an simple P4 (2.25) , 1gb ram, maudio 
audiophile card.

What Jack settings do you use?

\r