[PD] gripd glitches

[Michal Seta]

On 9/4/01 9:57 PM, "Joseph A. Sarlo" <jsarlo at mambo.peabody.jhu.edu> wrote:

> Well, you should definately run PD with -rt. If you're calling GrIPD with
> an "open" message, the GrIPD GUI will run with priority 0 (SCHED_OTHER)
> schedualing (as opposed to PD's 96), so that helps a lot. Are you getting
> DIO errors?  I've found in some of the later versions of PD that I
> sometimes get a crackling in audio with no DIO errors and low CPU usage,
> and I'm not sure why.

no DIO errors.  

> Can you run other reasonably low utilization apps
> simultaneously with your patch without problem or is it just GrIPD?

Yup.  Gimp causes far less interruptions.  Every time I move a mouse within
the bounds of GrIPD's window I get lots of clackling.  And I mean a lot.
Kinda like granular synthesis :)

> 
> SOCKETTIMERTIME = 5

poll_send/receive doesn't change anything but SOCKETTIMERTIME does but not a
lot.  And this really happen only when I read from arrays using tabread4~.
Any other patch works fine.


>  Also, I'm not sure, but using a non-bitmap GTK theme might help.
> I've also found, from purely a voodoo standpoint, that opening the GrIPD
> GUI before I start audio seems to help.

Yeah, non-bitmap GTK.  And I always opened GUI before audio....



However, from the little monitoring I did, I think my main problem is RAM.
I don't know how related that is but I've noticed that every time I move the
mouse around X(server) gets all the way up in 'top'.  Also, top reports that
X size is 119M which is almost twice the size of my RAM.  Which explains
%MEM = 194.  Assuming that arrays are sitting in the memory as well I'm led
to think that, indeed, RAM is at cause here.  I think I will give up GrIPD
until I get/borrow some RAM and will test it then again.  For now, I almost
finished my GUI with iemlib :)

Thanks anyways. 


./MiS