[PD] [metro] [realtime] and [cputime], were is the true?

Johannes M Zmoelnig zmoelnig at iem.at
Tue Nov 9 12:02:06 CET 2004

stephane wrote:
> The second :
> I don't understand were come the near 1.5 % error between [realtime] & 
> [metro] objects. It seam to be different from a computer to another. 
> Maybe a latency problem... mmh...?

well, messages are handled at a "single instant of time" each DSP-cycle 
(which is approx 1.5ms with 64samples @ 44.1kHz)
of course the "single instant" is not "at once" but takes some time 
itself (due to the nature of sequential CPUs).

[realtime] basically measures the number of CPU-ticks between 2 bangs.
running an operating system on your computer might use some ticks now 
and then; when your systems eats more ressources, the [metro] might be 
called a bit later, resulting in jitter.


More information about the Pd-list mailing list