[PD] Questions about PD speed

thewade pdman at aproximation.org
Mon Mar 6 03:52:40 CET 2006


Hello,

I am wondering about (1)the processor costs of [s ]/[r ] and [s~ ]/[r~ 
], and (2)if PD would benefit from running in a multiprocessor 
environment?

(1) How much processing power do I loose when I use [send~ ] and 
multiple [receive~ ] instead of just patching as normal? How about at 
the control rate with [send ]/[receive ]?

Do I have the following list correct - processor eaters ordered by most 
greedy at the top to least greedy at the bottom:
a. Arrays graphed on parent
b. general gui stuff like sliders/number-boxes graphed on parent
c. [s~]/[r~]
d. [s]/[r]

Do I loose extra of I have an object graphing on parent inside of 
another object graphing on parent (nested graphing objects), as opposed 
to just two graphing objects not nested? It seems like I do...

(2) I know that there is a threaded version of soundfiler, but are 
there any advantages to using PD on a multi-processor environment? Can 
you use to instantiations of PD on the same machine with two processors 
and get GEM and PD to play nice togeather (no popping during 
screen-redraw)? How about AMD64 (and are arrays working yet in 64-bit 
mode)?

Oh, BTW, I have a stream up of some of my music - the majority of which 
is made using PD: http://MegaDolkaDot.com:8000/listen.pls
I don't have much bandwidth so there is a max of 4 listeners... When I 
get some money coming in I will set up a live365 stream or something.

Thanks all!
-thewade





More information about the Pd-list mailing list