[PD] jack

[guenter geiger]

On Sat, 29 Mar 2003, Larry Troxler wrote:
> On Thursday 27 March 2003 20:42, Antonis Galanopoulos wrote:
> > Hi Guenter,
> > i got the latest cvs and it works fine. the problem with the moving objects
> > is gone although I didnt test that much. It hasn't crashed yet and it is
> > running for quite a while now .
> > I noticed however that the CPU usage has gone up. It was 2-3 % before and
> > it is 9-10 now with the same patch. I guess that has to do with the less
> > priority pd now has. Thats not a problem though, I'm just letting you know.
> > Thanks very much.
>
> A question for Guenter, or whoever else: how is the CPU usage measured?
> Is it the time at the start of a DSP cycle to the time at the end of the
> cycle, or is it really only the CPU time that the PD process actually uses,
> taking into account that PD could be pre-empted in the middle of the
> processing (I think that before jack, this was probably not an issue before)?


Hi,

On UNIX pd meaasures performance with the times() function. This should
give back the exact times the process needed. I do not know why the latest
changs should influence CPU usage. I haven't noticed something similar
until now.

Could be that connecting all theports by default makes a difference on
Antonis system. (Before only stereo output was connected)

Guenter

>
>
> If the latter (true CPU usage is measured), I don't understand it. Why would
> PD's CPU usage increase by that much?
>
> If the former (only the time between the start and end of the processing is
> measured) I suppose it means that PD is being pre-empted, presumably by jack.
> But why would this be happening, if PD is fast enough to complete its
> processing in time (and hence not get pre-empted by jack due to timeout)?
>
> Larry Troxler
>
>
>
>
>
> _______________________________________________
> PD-list mailing list
> PD-list at iem.kug.ac.at
> http://iem.kug.ac.at/cgi-bin/mailman/listinfo/pd-list
>